Artificial Intelligence Nanodegree

Computer Vision Capstone

Project: Facial Keypoint Detection

---

Welcome to the final Computer Vision project in the Artificial Intelligence Nanodegree program!

In this project, you’ll combine your knowledge of computer vision techniques and deep learning to build and end-to-end facial keypoint recognition system! Facial keypoints include points around the eyes, nose, and mouth on any face and are used in many applications, from facial tracking to emotion recognition.

There are three main parts to this project:

Part 1 : Investigating OpenCV, pre-processing, and face detection

Part 2 : Training a Convolutional Neural Network (CNN) to detect facial keypoints

Part 3 : Putting parts 1 and 2 together to identify facial keypoints on any image!

---

*Here's what you need to know to complete the project:

1. In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested.

   a. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

1. In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation.

   a. Each section where you will answer a question is preceded by a 'Question X' header.

   b. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional suggestions for enhancing the project beyond the minimum requirements. If you decide to pursue the "(Optional)" sections, you should include the code in this IPython notebook.

Your project submission will be evaluated based on your answers to each of the questions and the code implementations you provide.

Steps to Complete the Project

Each part of the notebook is further broken down into separate steps. Feel free to use the links below to navigate the notebook.

In this project you will get to explore a few of the many computer vision algorithms built into the OpenCV library. This expansive computer vision library is now almost 20 years old and still growing!

The project itself is broken down into three large parts, then even further into separate steps. Make sure to read through each step, and complete any sections that begin with '(IMPLEMENTATION)' in the header; these implementation sections may contain multiple TODOs that will be marked in code. For convenience, we provide links to each of these steps below.

Part 1 : Investigating OpenCV, pre-processing, and face detection

• Step 0: Detect Faces Using a Haar Cascade Classifier
• Step 1: Add Eye Detection
• Step 2: De-noise an Image for Better Face Detection
• Step 3: Blur an Image and Perform Edge Detection
• Step 4: Automatically Hide the Identity of an Individual

Part 2 : Training a Convolutional Neural Network (CNN) to detect facial keypoints

• Step 5: Create a CNN to Recognize Facial Keypoints
• Step 6: Compile and Train the Model
• Step 7: Visualize the Loss and Answer Questions

Part 3 : Putting parts 1 and 2 together to identify facial keypoints on any image!

• Step 8: Build a Robust Facial Keypoints Detector (Complete the CV Pipeline)

---

Step 0: Detect Faces Using a Haar Cascade Classifier

Have you ever wondered how Facebook automatically tags images with your friends' faces? Or how high-end cameras automatically find and focus on a certain person's face? Applications like these depend heavily on the machine learning task known as face detection - which is the task of automatically finding faces in images containing people.

At its root face detection is a classification problem - that is a problem of distinguishing between distinct classes of things. With face detection these distinct classes are 1) images of human faces and 2) everything else.

We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the detector_architectures directory.

Import Resources

In the next python cell, we load in the required libraries for this section of the project.

# Import required libraries for this section

%matplotlib inline

import numpy as np
import matplotlib.pyplot as plt
import math
import cv2                     # OpenCV library for computer vision
from PIL import Image
import time 

Next, we load in and display a test image for performing face detection.

Note: by default OpenCV assumes the ordering of our image's color channels are Blue, then Green, then Red. This is slightly out of order with most image types we'll use in these experiments, whose color channels are ordered Red, then Green, then Blue. In order to switch the Blue and Red channels of our test image around we will use OpenCV's cvtColor function, which you can read more about by checking out some of its documentation located here. This is a general utility function that can do other transformations too like converting a color image to grayscale, and transforming a standard color image to HSV color space.

# Load in color image for face detection
image = cv2.imread('images/test_image_1.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Plot our image using subplots to specify a size and title
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)

<matplotlib.image.AxesImage at 0x1f182b44550>

There are a lot of people - and faces - in this picture. 13 faces to be exact! In the next code cell, we demonstrate how to use a Haar Cascade classifier to detect all the faces in this test image.

This face detector uses information about patterns of intensity in an image to reliably detect faces under varying light conditions. So, to use this face detector, we'll first convert the image from color to grayscale.

Then, we load in the fully trained architecture of the face detector -- found in the file haarcascade_frontalface_default.xml - and use it on our image to find faces!

To learn more about the parameters of the detector see this post.

# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face Detections')
ax1.imshow(image_with_detections)

Number of faces detected: 13

<matplotlib.image.AxesImage at 0x1f1833714a8>

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

---

Step 1: Add Eye Detections

There are other pre-trained detectors available that use a Haar Cascade Classifier - including full human body detectors, license plate detectors, and more. A full list of the pre-trained architectures can be found here.

To test your eye detector, we'll first read in a new test image with just a single face.

# Load in color image for face detection
image = cv2.imread('images/james.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Plot the RGB image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)

<matplotlib.image.AxesImage at 0x1f181fbf3c8>

Notice that even though the image is a black and white image, we have read it in as a color image and so it will still need to be converted to grayscale in order to perform the most accurate face detection.

So, the next steps will be to convert this image to grayscale, then load OpenCV's face detector and run it with parameters that detect this face accurately.

# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 1.25, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face Detection')
ax1.imshow(image_with_detections)

Number of faces detected: 1

<matplotlib.image.AxesImage at 0x1f182329978>

(IMPLEMENTATION) Add an eye detector to the current face detection setup.

A Haar-cascade eye detector can be included in the same way that the face detector was and, in this first task, it will be your job to do just this.

To set up an eye detector, use the stored parameters of the eye cascade detector, called haarcascade_eye.xml, located in the detector_architectures subdirectory. In the next code cell, create your eye detector and store its detections.

A few notes before you get started:

First, make sure to give your loaded eye detector the variable name

eye_cascade

and give the list of eye regions you detect the variable name

eyes

Second, since we've already run the face detector over this image, you should only search for eyes within the rectangular face regions detected in faces. This will minimize false detections.

Lastly, once you've run your eye detector over the facial detection region, you should display the RGB image with both the face detection boxes (in red) and your eye detections (in green) to verify that everything works as expected.

# Make a copy of the original image to plot rectangle detections
image_with_detections = np.copy(image)   

# Loop over the detections and draw their corresponding face detection boxes
for (x,y,w,h) in faces:
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h),(255,0,0), 3)  
    
# Do not change the code above this comment!

    
## TODO: Add eye detection, using haarcascade_eye.xml, to the current face detector algorithm
## TODO: Loop over the eye detections and draw their corresponding boxes in green on image_with_detections
eye_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_eye.xml')

# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

for (x,y,w,h) in faces:
    image_face_rect = gray[y:y+h, x:x+w]
    eyes = eye_cascade.detectMultiScale(image_face_rect)
    
    for (x_e,y_e,w_e,h_e) in eyes:
        # Add a green bounding box to the detections image
        cv2.rectangle(image_with_detections, (x + x_e, y + y_e), (x + x_e + w_e, y + y_e + h_e), (0,255,0), 3)
    

# Plot the image with both faces and eyes detected
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face and Eye Detection')
ax1.imshow(image_with_detections)

<matplotlib.image.AxesImage at 0x1f182504b00>

(Optional) Add face and eye detection to your laptop camera

It's time to kick it up a notch, and add face and eye detection to your laptop's camera! Afterwards, you'll be able to show off your creation like in the gif shown below - made with a completed version of the code!

Notice that not all of the detections here are perfect - and your result need not be perfect either. You should spend a small amount of time tuning the parameters of your detectors to get reasonable results, but don't hold out for perfection. If we wanted perfection we'd need to spend a ton of time tuning the parameters of each detector, cleaning up the input image frames, etc. You can think of this as more of a rapid prototype.

The next cell contains code for a wrapper function called laptop_camera_face_eye_detector that, when called, will activate your laptop's camera. You will place the relevant face and eye detection code in this wrapper function to implement face/eye detection and mark those detections on each image frame that your camera captures.

Before adding anything to the function, you can run it to get an idea of how it works - a small window should pop up showing you the live feed from your camera; you can press any key to close this window.

Note: Mac users may find that activating this function kills the kernel of their notebook every once in a while. If this happens to you, just restart your notebook's kernel, activate cell(s) containing any crucial import statements, and you'll be good to go!

### Add face and eye detection to this laptop camera function 
# Make sure to draw out all faces/eyes found in each frame on the shown video feed

import cv2
import time 

# wrapper function for face/eye detection with your laptop camera
def laptop_camera_go():
    # Extract the pre-trained face and eye detectors from xml files
    face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')
    eye_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_eye.xml')

    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Define the codec and create VideoWriter object
    fourcc = cv2.VideoWriter_fourcc(*'XVID')
    out = cv2.VideoWriter('output.avi',fourcc, 10.0, (640,480))

    # Keep the video stream open
    while rval:
        
        # Convert the RGB  image to grayscale
        gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)

        # Detect the faces in image
        faces = face_cascade.detectMultiScale(gray, 1.1, 4)
        
        # Make a copy of the orginal image to draw face detections on
        image_with_detections = np.copy(frame)

        # Get the bounding box for each detected face
        for (x,y,w,h) in faces:
            image_face_rect = gray[y:y+h, x:x+w]
            eyes = eye_cascade.detectMultiScale(image_face_rect, 1.2, 6)            
            
            for (x_e,y_e,w_e,h_e) in eyes:
                # Add a green bounding box to the detections image
                cv2.rectangle(image_with_detections, (x + x_e, y + y_e), (x + x_e + w_e, y + y_e + h_e), (0,255,0), 3)            
        
            # Add a red bounding box to the detections image
            cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)

        # write the frame
        out.write(image_with_detections)
        
        # Plot the image from camera with all the face and eye detections marked
        cv2.imshow("face detection activated", image_with_detections)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # Exit by pressing any key
            vc.release()
            out.release()

            # Destroy windows 
            cv2.destroyAllWindows()
          
            # Make sure window closes on OSx
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    

# Call the laptop camera face/eye detector function above
laptop_camera_go()

---

Step 2: De-noise an Image for Better Face Detection

Image quality is an important aspect of any computer vision task. Typically, when creating a set of images to train a deep learning network, significant care is taken to ensure that training images are free of visual noise or artifacts that hinder object detection. While computer vision algorithms - like a face detector - are typically trained on 'nice' data such as this, new test data doesn't always look so nice!

When applying a trained computer vision algorithm to a new piece of test data one often cleans it up first before feeding it in. This sort of cleaning - referred to as pre-processing - can include a number of cleaning phases like blurring, de-noising, color transformations, etc., and many of these tasks can be accomplished using OpenCV.

In this short subsection we explore OpenCV's noise-removal functionality to see how we can clean up a noisy image, which we then feed into our trained face detector.

Create a noisy image to work with

In the next cell, we create an artificial noisy version of the previous multi-face image. This is a little exaggerated - we don't typically get images that are this noisy - but image noise, or 'grainy-ness' in a digitial image - is a fairly common phenomenon.

# Load in the multi-face test image again
image = cv2.imread('images/test_image_1.jpg')

# Convert the image copy to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Make an array copy of this image
image_with_noise = np.asarray(image)

# Create noise - here we add noise sampled randomly from a Gaussian distribution: a common model for noise
noise_level = 45
noise = np.random.randn(image.shape[0],image.shape[1],image.shape[2])*noise_level

# Add this noise to the array image copy
image_with_noise = image_with_noise + noise

# Convert back to uint8 format
image_with_noise = np.asarray([np.uint8(np.clip(i,0,255)) for i in image_with_noise])

# Plot our noisy image!
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Noisy Image')
ax1.imshow(image_with_noise)

<matplotlib.image.AxesImage at 0x1f18460f7b8>

In the context of face detection, the problem with an image like this is that - due to noise - we may miss some faces or get false detections.

In the next cell we apply the same trained OpenCV detector with the same settings as before, to see what sort of detections we get.

# Convert the RGB  image to grayscale
gray_noise = cv2.cvtColor(image_with_noise, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray_noise, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image_with_noise)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Noisy Image with Face Detections')
ax1.imshow(image_with_detections)

Number of faces detected: 12

<matplotlib.image.AxesImage at 0x1f1822e5630>

With this added noise we now miss one of the faces!

(IMPLEMENTATION) De-noise this image for better face detection

Time to get your hands dirty: using OpenCV's built in color image de-noising functionality called fastNlMeansDenoisingColored - de-noise this image enough so that all the faces in the image are properly detected. Once you have cleaned the image in the next cell, use the cell that follows to run our trained face detector over the cleaned image to check out its detections.

You can find its official documentation here and a useful example here.

Note: you can keep all parameters except photo_render fixed as shown in the second link above. Play around with the value of this parameter - see how it affects the resulting cleaned image.

## TODO: Use OpenCV's built in color image de-noising function to clean up our noisy image!


denoised_image = cv2.fastNlMeansDenoisingColored(image_with_noise, None, 20, 20, 7, 21)

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Denoised Image')
ax1.imshow(denoised_image)

<matplotlib.image.AxesImage at 0x1f184680518>

## TODO: Run the face detector on the de-noised image to improve your detections and display the result
# Convert the RGB  image to grayscale
gray_denoised = cv2.cvtColor(denoised_image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray_denoised, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(denoised_image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Denoised Image with Face Detections')
ax1.imshow(image_with_detections)

Number of faces detected: 12

<matplotlib.image.AxesImage at 0x1f184717940>

---

Step 3: Blur an Image and Perform Edge Detection

Now that we have developed a simple pipeline for detecting faces using OpenCV - let's start playing around with a few fun things we can do with all those detected faces!

Importance of Blur in Edge Detection

Edge detection is a concept that pops up almost everywhere in computer vision applications, as edge-based features (as well as features built on top of edges) are often some of the best features for e.g., object detection and recognition problems.

Edge detection is a dimension reduction technique - by keeping only the edges of an image we get to throw away a lot of non-discriminating information. And typically the most useful kind of edge-detection is one that preserves only the important, global structures (ignoring local structures that aren't very discriminative). So removing local structures / retaining global structures is a crucial pre-processing step to performing edge detection in an image, and blurring can do just that.

Below is an animated gif showing the result of an edge-detected cat taken from Wikipedia, where the image is gradually blurred more and more prior to edge detection. When the animation begins you can't quite make out what it's a picture of, but as the animation evolves and local structures are removed via blurring the cat becomes visible in the edge-detected image.

Edge detection is a convolution performed on the image itself, and you can read about Canny edge detection on this OpenCV documentation page.

Canny edge detection

In the cell below we load in a test image, then apply Canny edge detection on it. The original image is shown on the left panel of the figure, while the edge-detected version of the image is shown on the right. Notice how the result looks very busy - there are too many little details preserved in the image before it is sent to the edge detector. When applied in computer vision applications, edge detection should preserve global structure; doing away with local structures that don't help describe what objects are in the image.

# Load in the image
image = cv2.imread('images/fawzia.jpg')

# Convert to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)  

# Perform Canny edge detection
edges = cv2.Canny(gray,100,200)

# Dilate the image to amplify edges
edges = cv2.dilate(edges, None)

# Plot the RGB and edge-detected image
fig = plt.figure(figsize = (15,15))
ax1 = fig.add_subplot(121)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)

ax2 = fig.add_subplot(122)
ax2.set_xticks([])
ax2.set_yticks([])

ax2.set_title('Canny Edges')
ax2.imshow(edges, cmap='gray')

<matplotlib.image.AxesImage at 0x1f1875dbd30>

Without first blurring the image, and removing small, local structures, a lot of irrelevant edge content gets picked up and amplified by the detector (as shown in the right panel above).

(IMPLEMENTATION) Blur the image then perform edge detection

In the next cell, you will repeat this experiment - blurring the image first to remove these local structures, so that only the important boudnary details remain in the edge-detected image.

Blur the image by using OpenCV's filter2d functionality - which is discussed in this documentation page - and use an averaging kernel of width equal to 4.

### TODO: Blur the test imageusing OpenCV's filter2d functionality, 
# Use an averaging kernel, and a kernel width equal to 4
# Convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)  

kernel = np.ones((4,4), np.float32) / 16
image_blurred = cv2.filter2D(gray, -1, kernel)
    
## TODO: Then perform Canny edge detection and display the output
edges = cv2.Canny(image_blurred,100,200)

# Dilate the image to amplify edges
edges = cv2.dilate(edges, None)

# Plot the RGB and edge-detected image
fig = plt.figure(figsize = (15,15))
ax1 = fig.add_subplot(121)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)

ax2 = fig.add_subplot(122)
ax2.set_xticks([])
ax2.set_yticks([])

ax2.set_title('Canny Edges')
ax2.imshow(edges, cmap='gray')

<matplotlib.image.AxesImage at 0x1f187684c88>

---

Step 4: Automatically Hide the Identity of an Individual

If you film something like a documentary or reality TV, you must get permission from every individual shown on film before you can show their face, otherwise you need to blur it out - by blurring the face a lot (so much so that even the global structures are obscured)! This is also true for projects like Google's StreetView maps - an enormous collection of mapping images taken from a fleet of Google vehicles. Because it would be impossible for Google to get the permission of every single person accidentally captured in one of these images they blur out everyone's faces, the detected images must automatically blur the identity of detected people. Here's a few examples of folks caught in the camera of a Google street view vehicle.

Read in an image to perform identity detection

Let's try this out for ourselves. Use the face detection pipeline built above and what you know about using the filter2D to blur and image, and use these in tandem to hide the identity of the person in the following image - loaded in and printed in the next cell.

# Load in the image
image = cv2.imread('images/gus.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Display the image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('Original Image')
ax1.imshow(image)

<matplotlib.image.AxesImage at 0x1f1871fe080>

(IMPLEMENTATION) Use blurring to hide the identity of an individual in an image

The idea here is to 1) automatically detect the face in this image, and then 2) blur it out! Make sure to adjust the parameters of the averaging blur filter to completely obscure this person's identity.

## TODO: Implement face detection
# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 1.3, 5)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image)

## TODO: Blur the bounding box around each detected face using an averaging filter and display the result
kernel = np.ones((100,100), np.float32) / 10000
# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    image_with_detections[y:y+h, x:x+w] = cv2.filter2D(image_with_detections[y:y+h, x:x+w], -1, kernel)

# Display the image with the detections
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face Detection')
ax1.imshow(image_with_detections)

Number of faces detected: 1

<matplotlib.image.AxesImage at 0x1f1861f2828>

(Optional) Build identity protection into your laptop camera

In this optional task you can add identity protection to your laptop camera, using the previously completed code where you added face detection to your laptop camera - and the task above. You should be able to get reasonable results with little parameter tuning - like the one shown in the gif below.

As with the previous video task, to make this perfect would require significant effort - so don't strive for perfection here, strive for reasonable quality.

The next cell contains code a wrapper function called laptop_camera_identity_hider that - when called - will activate your laptop's camera. You need to place the relevant face detection and blurring code developed above in this function in order to blur faces entering your laptop camera's field of view.

Before adding anything to the function you can call it to get a hang of how it works - a small window will pop up showing you the live feed from your camera, you can press any key to close this window.

Note: Mac users may find that activating this function kills the kernel of their notebook every once in a while. If this happens to you, just restart your notebook's kernel, activate cell(s) containing any crucial import statements, and you'll be good to go!

### Insert face detection and blurring code into the wrapper below to create an identity protector on your laptop!
import cv2
import time 

def laptop_camera_go():
    # Extract the pre-trained face and eye detectors from xml files
    face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')
    
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Define the codec and create VideoWriter object
    fourcc = cv2.VideoWriter_fourcc(*'XVID')
    out = cv2.VideoWriter('output.avi',fourcc, 10.0, (640,480))
    
    # Keep video stream open
    while rval:
        # Convert the RGB  image to grayscale
        gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)

        # Detect the faces in image
        faces = face_cascade.detectMultiScale(gray, 1.1, 4)
        
        # Make a copy of the orginal image to draw face detections on
        image_with_detections = np.copy(frame)

        # Get the bounding box for each detected face
        for (x,y,w,h) in faces:
            image_face_rect = gray[y:y+h, x:x+w]

            kernel = np.ones((50,50), np.float32) / 2500
            image_with_detections[y:y+h, x:x+w] = cv2.filter2D(image_with_detections[y:y+h, x:x+w], -1, kernel)
        
        # write the frame
        out.write(image_with_detections)

        # Plot the image from camera with all the face and eye detections marked
        cv2.imshow("face detection activated", image_with_detections)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # Exit by pressing any key
            vc.release()
            out.release()

            # Destroy windows
            cv2.destroyAllWindows()
            
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    
        

# Run laptop identity hider
laptop_camera_go()

---

Step 5: Create a CNN to Recognize Facial Keypoints

OpenCV is often used in practice with other machine learning and deep learning libraries to produce interesting results. In this stage of the project you will create your own end-to-end pipeline - employing convolutional networks in keras along with OpenCV - to apply a "selfie" filter to streaming video and images.

You will start by creating and then training a convolutional network that can detect facial keypoints in a small dataset of cropped images of human faces. We then guide you towards OpenCV to expanding your detection algorithm to more general images. What are facial keypoints? Let's take a look at some examples.

Facial keypoints (also called facial landmarks) are the small blue-green dots shown on each of the faces in the image above - there are 15 keypoints marked in each image. They mark important areas of the face - the eyes, corners of the mouth, the nose, etc. Facial keypoints can be used in a variety of machine learning applications from face and emotion recognition to commercial applications like the image filters popularized by Snapchat.

Below we illustrate a filter that, using the results of this section, automatically places sunglasses on people in images (using the facial keypoints to place the glasses correctly on each face). Here, the facial keypoints have been colored lime green for visualization purposes.

Make a facial keypoint detector

But first things first: how can we make a facial keypoint detector? Well, at a high level, notice that facial keypoint detection is a regression problem. A single face corresponds to a set of 15 facial keypoints (a set of 15 corresponding $(x, y)$ coordinates, i.e., an output point). Because our input data are images, we can employ a convolutional neural network to recognize patterns in our images and learn how to identify these keypoint given sets of labeled data.

In order to train a regressor, we need a training set - a set of facial image / facial keypoint pairs to train on. For this we will be using this dataset from Kaggle. We've already downloaded this data and placed it in the data directory. Make sure that you have both the training and test data files. The training dataset contains several thousand $96 \times 96$ grayscale images of cropped human faces, along with each face's 15 corresponding facial keypoints (also called landmarks) that have been placed by hand, and recorded in $(x, y)$ coordinates. This wonderful resource also has a substantial testing set, which we will use in tinkering with our convolutional network.

To load in this data, run the Python cell below - notice we will load in both the training and testing sets.

The load_data function is in the included utils.py file.

from utils import *

# Load training set
X_train, y_train = load_data()
print("X_train.shape == {}".format(X_train.shape))
print("y_train.shape == {}; y_train.min == {:.3f}; y_train.max == {:.3f}".format(
    y_train.shape, y_train.min(), y_train.max()))

# Load testing set
X_test, _ = load_data(test=True)
print("X_test.shape == {}".format(X_test.shape))

Using TensorFlow backend.

X_train.shape == (2140, 96, 96, 1)
y_train.shape == (2140, 30); y_train.min == -0.920; y_train.max == 0.996
X_test.shape == (1783, 96, 96, 1)

The load_data function in utils.py originates from this excellent blog post, which you are strongly encouraged to read. Please take the time now to review this function. Note how the output values - that is, the coordinates of each set of facial landmarks - have been normalized to take on values in the range $[-1, 1]$, while the pixel values of each input point (a facial image) have been normalized to the range $[0,1]$.

Note: the original Kaggle dataset contains some images with several missing keypoints. For simplicity, the load_data function removes those images with missing labels from the dataset. As an optional extension, you are welcome to amend the load_data function to include the incomplete data points.

Visualize the Training Data

Execute the code cell below to visualize a subset of the training data.

import matplotlib.pyplot as plt
%matplotlib inline

fig = plt.figure(figsize=(20,20))
fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)
for i in range(9):
    ax = fig.add_subplot(3, 3, i + 1, xticks=[], yticks=[])
    plot_data(X_train[i], y_train[i], ax)

import matplotlib.pyplot as plt
%matplotlib inline

flip_indices = [
        (0, 2), (1, 3),
        (4, 8), (5, 9), (6, 10), (7, 11),
        (12, 16), (13, 17), (14, 18), (15, 19),
        (22, 24), (23, 25),
        ]

fig = plt.figure(figsize=(5,5))
fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)

i = 1
ax = fig.add_subplot(1, 1, 1, xticks=[], yticks=[])
X_flipped = X_train[i, :, ::-1, :]  # simple slice to flip all images

print(y_train[i])

y_flipped = np.zeros(y_train.shape[1])

y_flipped[1::2] = y_train[i, 1::2]   
y_flipped[::2] = y_train[i, ::2] * -1.0

print(y_flipped)

# Swap places, e.g. left_eye_center_x -> right_eye_center_x
for a, b in flip_indices:
    y_flipped[a], y_flipped[b] = (
        y_flipped[b], y_flipped[a])

plot_data(X_flipped, y_flipped, ax)

[ 0.4330242  -0.21624877 -0.34668279 -0.3463223   0.25858903 -0.1851669
  0.58100623 -0.1878271  -0.16970447 -0.23996718 -0.5165956  -0.35648429
  0.29375893 -0.24669313  0.74820185 -0.32774875 -0.07298451 -0.29718679
 -0.62243527 -0.51643556 -0.0278047   0.24908654  0.33497101  0.3931978
 -0.4643302   0.31000873 -0.06167792  0.52398473 -0.08612007  0.59259433]
[-0.4330242  -0.21624877  0.34668279 -0.3463223  -0.25858903 -0.1851669
 -0.58100623 -0.1878271   0.16970447 -0.23996718  0.5165956  -0.35648429
 -0.29375893 -0.24669313 -0.74820185 -0.32774875  0.07298451 -0.29718679
  0.62243527 -0.51643556  0.0278047   0.24908654 -0.33497101  0.3931978
  0.4643302   0.31000873  0.06167792  0.52398473  0.08612007  0.59259433]

For each training image, there are two landmarks per eyebrow (four total), three per eye (six total), four for the mouth, and one for the tip of the nose.

Review the plot_data function in utils.py to understand how the 30-dimensional training labels in y_train are mapped to facial locations, as this function will prove useful for your pipeline.

(IMPLEMENTATION) Specify the CNN Architecture

In this section, you will specify a neural network for predicting the locations of facial keypoints. Use the code cell below to specify the architecture of your neural network. We have imported some layers that you may find useful for this task, but if you need to use more Keras layers, feel free to import them in the cell.

Your network should accept a $96 \times 96$ grayscale image as input, and it should output a vector with 30 entries, corresponding to the predicted (horizontal and vertical) locations of 15 facial keypoints. If you are not sure where to start, you can find some useful starting architectures in this blog, but you are not permitted to copy any of the architectures that you find online.

# CNN architecture for training with no augmentation

# Import deep learning resources from Keras
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Dropout, Activation
from keras.layers import Flatten, Dense


## TODO: Specify a CNN architecture
# Your model should accept 96x96 pixel graysale images in
# It should have a fully-connected output layer with 30 values (2 for each facial keypoint)

modelNoAugm = Sequential()

modelNoAugm.add(Conv2D(filters=32, kernel_size=3, padding='same', activation='relu', 
                        input_shape=(96, 96, 1)))
modelNoAugm.add(Conv2D(filters=32, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(MaxPooling2D(pool_size=2))
modelNoAugm.add(Dropout(0.3))

modelNoAugm.add(Conv2D(filters=64, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(Conv2D(filters=64, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(MaxPooling2D(pool_size=2))
modelNoAugm.add(Dropout(0.3))

modelNoAugm.add(Conv2D(filters=128, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(Conv2D(filters=128, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(MaxPooling2D(pool_size=2))
modelNoAugm.add(Dropout(0.3))

modelNoAugm.add(Conv2D(filters=256, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(Conv2D(filters=256, kernel_size=3, padding='same', activation='relu'))
modelNoAugm.add(MaxPooling2D(pool_size=2))
modelNoAugm.add(Dropout(0.3))

modelNoAugm.add(Flatten())

modelNoAugm.add(Dense(500))
modelNoAugm.add(Activation('relu'))
modelNoAugm.add(Dropout(0.4))

modelNoAugm.add(Dense(30))

# Summarize the model
modelNoAugm.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_9 (Conv2D)            (None, 96, 96, 32)        320       
_________________________________________________________________
conv2d_10 (Conv2D)           (None, 96, 96, 32)        9248      
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 48, 48, 32)        0         
_________________________________________________________________
dropout_6 (Dropout)          (None, 48, 48, 32)        0         
_________________________________________________________________
conv2d_11 (Conv2D)           (None, 48, 48, 64)        18496     
_________________________________________________________________
conv2d_12 (Conv2D)           (None, 48, 48, 64)        36928     
_________________________________________________________________
max_pooling2d_6 (MaxPooling2 (None, 24, 24, 64)        0         
_________________________________________________________________
dropout_7 (Dropout)          (None, 24, 24, 64)        0         
_________________________________________________________________
conv2d_13 (Conv2D)           (None, 24, 24, 128)       73856     
_________________________________________________________________
conv2d_14 (Conv2D)           (None, 24, 24, 128)       147584    
_________________________________________________________________
max_pooling2d_7 (MaxPooling2 (None, 12, 12, 128)       0         
_________________________________________________________________
dropout_8 (Dropout)          (None, 12, 12, 128)       0         
_________________________________________________________________
conv2d_15 (Conv2D)           (None, 12, 12, 256)       295168    
_________________________________________________________________
conv2d_16 (Conv2D)           (None, 12, 12, 256)       590080    
_________________________________________________________________
max_pooling2d_8 (MaxPooling2 (None, 6, 6, 256)         0         
_________________________________________________________________
dropout_9 (Dropout)          (None, 6, 6, 256)         0         
_________________________________________________________________
flatten_2 (Flatten)          (None, 9216)              0         
_________________________________________________________________
dense_3 (Dense)              (None, 500)               4608500   
_________________________________________________________________
activation_2 (Activation)    (None, 500)               0         
_________________________________________________________________
dropout_10 (Dropout)         (None, 500)               0         
_________________________________________________________________
dense_4 (Dense)              (None, 30)                15030     
=================================================================
Total params: 5,795,210
Trainable params: 5,795,210
Non-trainable params: 0
_________________________________________________________________

# CNN architecture for training on augmented data

# Import deep learning resources from Keras
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Dropout, Activation
from keras.layers import Flatten, Dense


## TODO: Specify a CNN architecture
# Your model should accept 96x96 pixel graysale images in
# It should have a fully-connected output layer with 30 values (2 for each facial keypoint)

model = Sequential()

model.add(Conv2D(filters=32, kernel_size=3, padding='same', activation='relu', 
                        input_shape=(96, 96, 1)))
model.add(Conv2D(filters=32, kernel_size=3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.3))

model.add(Conv2D(filters=64, kernel_size=3, padding='same', activation='relu'))
model.add(Conv2D(filters=64, kernel_size=3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.3))

model.add(Conv2D(filters=128, kernel_size=3, padding='same', activation='relu'))
model.add(Conv2D(filters=128, kernel_size=3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.4))

model.add(Conv2D(filters=256, kernel_size=3, padding='same', activation='relu'))
model.add(Conv2D(filters=256, kernel_size=3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.4))

model.add(Flatten())

model.add(Dense(500))
model.add(Activation('relu'))
model.add(Dropout(0.4))

model.add(Dense(30))

# Summarize the model
model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 96, 96, 32)        320       
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 96, 96, 32)        9248      
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 48, 48, 32)        0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 48, 48, 32)        0         
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 48, 48, 64)        18496     
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 48, 48, 64)        36928     
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 24, 24, 64)        0         
_________________________________________________________________
dropout_2 (Dropout)          (None, 24, 24, 64)        0         
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 24, 24, 128)       73856     
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 24, 24, 128)       147584    
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 12, 12, 128)       0         
_________________________________________________________________
dropout_3 (Dropout)          (None, 12, 12, 128)       0         
_________________________________________________________________
conv2d_7 (Conv2D)            (None, 12, 12, 256)       295168    
_________________________________________________________________
conv2d_8 (Conv2D)            (None, 12, 12, 256)       590080    
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 6, 6, 256)         0         
_________________________________________________________________
dropout_4 (Dropout)          (None, 6, 6, 256)         0         
_________________________________________________________________
flatten_1 (Flatten)          (None, 9216)              0         
_________________________________________________________________
dense_1 (Dense)              (None, 500)               4608500   
_________________________________________________________________
activation_1 (Activation)    (None, 500)               0         
_________________________________________________________________
dropout_5 (Dropout)          (None, 500)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 30)                15030     
=================================================================
Total params: 5,795,210
Trainable params: 5,795,210
Non-trainable params: 0
_________________________________________________________________

---

Step 6: Compile and Train the Model

After specifying your architecture, you'll need to compile and train the model to detect facial keypoints'

(IMPLEMENTATION) Compile and Train the Model

Use the compile method to configure the learning process. Experiment with your choice of optimizer; you may have some ideas about which will work best (SGD vs. RMSprop, etc), but take the time to empirically verify your theories.

Use the fit method to train the model. Break off a validation set by setting validation_split=0.2. Save the returned History object in the history variable.

Your model is required to attain a validation loss (measured as mean squared error) of at least XYZ. When you have finished training, save your model as an HDF5 file with file path my_model.h5.

# Get the validation set size (20%)
div = X_train.shape[0] // 5

# Split the data set to train and validation sets 
X_train_d, X_val = np.split(X_train, [X_train.shape[0] - div])
y_train_d, y_val = np.split(y_train, [y_train.shape[0] - div])


print("Train set X,y shapes: {}:{}".format(X_train_d.shape, y_train_d.shape))
print("Validation set X,y shapes: {}:{}".format(X_val.shape, y_val.shape))

Train set X,y shapes: (1712, 96, 96, 1):(1712, 30)
Validation set X,y shapes: (428, 96, 96, 1):(428, 30)

# Training without augmentation  

from keras.optimizers import SGD, RMSprop, Adagrad, Adadelta, Adam, Adamax, Nadam
from keras.callbacks import ModelCheckpoint

## TODO: Compile the model
modelNoAugm.compile(optimizer='adam', loss='mean_squared_error')

## TODO: Train the model

checkpointer = ModelCheckpoint(filepath='my_model.h5',
                               verbose=1, save_best_only=True)
  

hist = modelNoAugm.fit(X_train, y_train, 
          epochs=1000, batch_size=214, validation_split=0.2, callbacks=[checkpointer], verbose=2)

## TODO: Save the model as model.h5
modelNoAugm.load_weights('my_model.h5')

# Training with augmentation  

from keras.optimizers import SGD, RMSprop, Adagrad, Adadelta, Adam, Adamax, Nadam
from keras.callbacks import ModelCheckpoint

BATCH_SIZE = 214
EPOCHS = 1500

## TODO: Compile the model
model.compile(optimizer='adam', loss='mean_squared_error')

## TODO: Train the model

checkpointer = ModelCheckpoint(filepath='my_model.h5',
                               verbose=1, save_best_only=True)
  
flip_indices = [
        (0, 2), (1, 3),
        (4, 8), (5, 9), (6, 10), (7, 11),
        (12, 16), (13, 17), (14, 18), (15, 19),
        (22, 24), (23, 25),
        ]


# Generator function for augmented images 
def generator_train(X, y, batch_size):
    batch_X = np.zeros((batch_size, X.shape[1], X.shape[2], X.shape[3]))
    batch_y = np.zeros((batch_size, y.shape[1]))
    
    while True:
        batch_indices = np.random.choice(X.shape[0], batch_size, replace=False)
        batch_X[:, :, :, :] = X[batch_indices, :, :, :]
        
        # Flip half of the images
        indices = np.random.choice(batch_size, batch_size // 2, replace=False)
        batch_X[indices] = batch_X[indices, :, ::-1, :]

        batch_y[:, :] = y[batch_indices, :]
        
        # Flip x coordinates of the points:
        batch_y[indices, ::2] = batch_y[indices, ::2] * -1
        
        # Swap point places
        for a, b in flip_indices:
            batch_y[indices, a], batch_y[indices, b] = (
                batch_y[indices, b], batch_y[indices, a])

        yield batch_X, batch_y


hist = model.fit_generator(generator_train(X_train_d, y_train_d, batch_size=BATCH_SIZE),
                    steps_per_epoch=X_train_d.shape[0] // BATCH_SIZE,
                    epochs=EPOCHS, verbose=2, callbacks=[checkpointer],
                    validation_data=(X_val, y_val))




## TODO: Save the model as model.h5
model.load_weights('my_model.h5')

Epoch 1/1500
Epoch 00000: val_loss improved from inf to 0.05940, saving model to my_model.h5
7s - loss: 0.0811 - val_loss: 0.0594
Epoch 2/1500
Epoch 00001: val_loss improved from 0.05940 to 0.01382, saving model to my_model.h5
3s - loss: 0.0203 - val_loss: 0.0138
Epoch 3/1500
Epoch 00002: val_loss improved from 0.01382 to 0.00869, saving model to my_model.h5
3s - loss: 0.0132 - val_loss: 0.0087
Epoch 4/1500
Epoch 00003: val_loss improved from 0.00869 to 0.00772, saving model to my_model.h5
3s - loss: 0.0110 - val_loss: 0.0077
Epoch 5/1500
Epoch 00004: val_loss improved from 0.00772 to 0.00653, saving model to my_model.h5
3s - loss: 0.0098 - val_loss: 0.0065
Epoch 6/1500
Epoch 00005: val_loss improved from 0.00653 to 0.00557, saving model to my_model.h5
3s - loss: 0.0090 - val_loss: 0.0056
Epoch 7/1500
Epoch 00006: val_loss improved from 0.00557 to 0.00492, saving model to my_model.h5
3s - loss: 0.0086 - val_loss: 0.0049
Epoch 8/1500
Epoch 00007: val_loss did not improve
3s - loss: 0.0082 - val_loss: 0.0053
Epoch 9/1500
Epoch 00008: val_loss did not improve
3s - loss: 0.0079 - val_loss: 0.0054
Epoch 10/1500
Epoch 00009: val_loss did not improve
3s - loss: 0.0076 - val_loss: 0.0052
Epoch 11/1500
Epoch 00010: val_loss improved from 0.00492 to 0.00461, saving model to my_model.h5
3s - loss: 0.0076 - val_loss: 0.0046
Epoch 12/1500
Epoch 00011: val_loss did not improve
3s - loss: 0.0075 - val_loss: 0.0048
Epoch 13/1500
Epoch 00012: val_loss did not improve
3s - loss: 0.0072 - val_loss: 0.0051
Epoch 14/1500
Epoch 00013: val_loss did not improve
3s - loss: 0.0071 - val_loss: 0.0051
Epoch 15/1500
Epoch 00014: val_loss did not improve
3s - loss: 0.0071 - val_loss: 0.0046
Epoch 16/1500
Epoch 00015: val_loss did not improve
3s - loss: 0.0069 - val_loss: 0.0050
Epoch 17/1500
Epoch 00016: val_loss did not improve
3s - loss: 0.0066 - val_loss: 0.0049
Epoch 18/1500
Epoch 00017: val_loss did not improve
3s - loss: 0.0066 - val_loss: 0.0047
Epoch 19/1500
Epoch 00018: val_loss did not improve
3s - loss: 0.0064 - val_loss: 0.0052
Epoch 20/1500
Epoch 00019: val_loss did not improve
3s - loss: 0.0066 - val_loss: 0.0049
Epoch 21/1500
Epoch 00020: val_loss did not improve
3s - loss: 0.0064 - val_loss: 0.0047
Epoch 22/1500
Epoch 00021: val_loss did not improve
3s - loss: 0.0065 - val_loss: 0.0049
Epoch 23/1500
Epoch 00022: val_loss did not improve
3s - loss: 0.0062 - val_loss: 0.0046
Epoch 24/1500
Epoch 00023: val_loss did not improve
3s - loss: 0.0061 - val_loss: 0.0051
Epoch 25/1500
Epoch 00024: val_loss did not improve
3s - loss: 0.0061 - val_loss: 0.0047
Epoch 26/1500
Epoch 00025: val_loss improved from 0.00461 to 0.00449, saving model to my_model.h5
3s - loss: 0.0063 - val_loss: 0.0045
Epoch 27/1500
Epoch 00026: val_loss did not improve
3s - loss: 0.0064 - val_loss: 0.0060
Epoch 28/1500
Epoch 00027: val_loss did not improve
3s - loss: 0.0062 - val_loss: 0.0048
Epoch 29/1500
Epoch 00028: val_loss did not improve
3s - loss: 0.0060 - val_loss: 0.0047
Epoch 30/1500
Epoch 00029: val_loss improved from 0.00449 to 0.00449, saving model to my_model.h5
3s - loss: 0.0059 - val_loss: 0.0045
Epoch 31/1500
Epoch 00030: val_loss did not improve
3s - loss: 0.0059 - val_loss: 0.0050
Epoch 32/1500
Epoch 00031: val_loss did not improve
3s - loss: 0.0060 - val_loss: 0.0047
Epoch 33/1500
Epoch 00032: val_loss did not improve
3s - loss: 0.0061 - val_loss: 0.0050
Epoch 34/1500
Epoch 00033: val_loss improved from 0.00449 to 0.00448, saving model to my_model.h5
3s - loss: 0.0057 - val_loss: 0.0045
Epoch 35/1500
Epoch 00034: val_loss did not improve
3s - loss: 0.0057 - val_loss: 0.0055
Epoch 36/1500
Epoch 00035: val_loss did not improve
3s - loss: 0.0059 - val_loss: 0.0045
Epoch 37/1500
Epoch 00036: val_loss did not improve
3s - loss: 0.0057 - val_loss: 0.0048
Epoch 38/1500
Epoch 00037: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0047
Epoch 39/1500
Epoch 00038: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0045
Epoch 40/1500
Epoch 00039: val_loss did not improve
3s - loss: 0.0058 - val_loss: 0.0045
Epoch 41/1500
Epoch 00040: val_loss did not improve
3s - loss: 0.0058 - val_loss: 0.0048
Epoch 42/1500
Epoch 00041: val_loss did not improve
3s - loss: 0.0059 - val_loss: 0.0045
Epoch 43/1500
Epoch 00042: val_loss improved from 0.00448 to 0.00437, saving model to my_model.h5
3s - loss: 0.0057 - val_loss: 0.0044
Epoch 44/1500
Epoch 00043: val_loss did not improve
3s - loss: 0.0055 - val_loss: 0.0049
Epoch 45/1500
Epoch 00044: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0044
Epoch 46/1500
Epoch 00045: val_loss did not improve
3s - loss: 0.0057 - val_loss: 0.0047
Epoch 47/1500
Epoch 00046: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0044
Epoch 48/1500
Epoch 00047: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0045
Epoch 49/1500
Epoch 00048: val_loss did not improve
3s - loss: 0.0055 - val_loss: 0.0047
Epoch 50/1500
Epoch 00049: val_loss did not improve
3s - loss: 0.0053 - val_loss: 0.0045
Epoch 51/1500
Epoch 00050: val_loss did not improve
3s - loss: 0.0056 - val_loss: 0.0049
Epoch 52/1500
Epoch 00051: val_loss did not improve
3s - loss: 0.0053 - val_loss: 0.0047
Epoch 53/1500
Epoch 00052: val_loss did not improve
3s - loss: 0.0055 - val_loss: 0.0047
Epoch 54/1500
Epoch 00053: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0047
Epoch 55/1500
Epoch 00054: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0047
Epoch 56/1500
Epoch 00055: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0045
Epoch 57/1500
Epoch 00056: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0052
Epoch 58/1500
Epoch 00057: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0046
Epoch 59/1500
Epoch 00058: val_loss improved from 0.00437 to 0.00432, saving model to my_model.h5
3s - loss: 0.0052 - val_loss: 0.0043
Epoch 60/1500
Epoch 00059: val_loss did not improve
3s - loss: 0.0054 - val_loss: 0.0045
Epoch 61/1500
Epoch 00060: val_loss did not improve
3s - loss: 0.0051 - val_loss: 0.0053
Epoch 62/1500
Epoch 00061: val_loss improved from 0.00432 to 0.00403, saving model to my_model.h5
3s - loss: 0.0052 - val_loss: 0.0040
Epoch 63/1500
Epoch 00062: val_loss did not improve
3s - loss: 0.0052 - val_loss: 0.0041
Epoch 64/1500
Epoch 00063: val_loss improved from 0.00403 to 0.00399, saving model to my_model.h5
3s - loss: 0.0051 - val_loss: 0.0040
Epoch 65/1500
Epoch 00064: val_loss improved from 0.00399 to 0.00393, saving model to my_model.h5
3s - loss: 0.0049 - val_loss: 0.0039
Epoch 66/1500
Epoch 00065: val_loss did not improve
3s - loss: 0.0050 - val_loss: 0.0046
Epoch 67/1500
Epoch 00066: val_loss did not improve
3s - loss: 0.0049 - val_loss: 0.0040
Epoch 68/1500
Epoch 00067: val_loss improved from 0.00393 to 0.00380, saving model to my_model.h5
3s - loss: 0.0048 - val_loss: 0.0038
Epoch 69/1500
Epoch 00068: val_loss did not improve
3s - loss: 0.0049 - val_loss: 0.0043
Epoch 70/1500
Epoch 00069: val_loss did not improve
3s - loss: 0.0047 - val_loss: 0.0039
Epoch 71/1500
Epoch 00070: val_loss improved from 0.00380 to 0.00353, saving model to my_model.h5
3s - loss: 0.0044 - val_loss: 0.0035
Epoch 72/1500
Epoch 00071: val_loss did not improve
3s - loss: 0.0045 - val_loss: 0.0036
Epoch 73/1500
Epoch 00072: val_loss did not improve
3s - loss: 0.0045 - val_loss: 0.0040
Epoch 74/1500
Epoch 00073: val_loss improved from 0.00353 to 0.00315, saving model to my_model.h5
3s - loss: 0.0043 - val_loss: 0.0031
Epoch 75/1500
Epoch 00074: val_loss improved from 0.00315 to 0.00306, saving model to my_model.h5
3s - loss: 0.0041 - val_loss: 0.0031
Epoch 76/1500
Epoch 00075: val_loss improved from 0.00306 to 0.00286, saving model to my_model.h5
3s - loss: 0.0039 - val_loss: 0.0029
Epoch 77/1500
Epoch 00076: val_loss improved from 0.00286 to 0.00279, saving model to my_model.h5
3s - loss: 0.0039 - val_loss: 0.0028
Epoch 78/1500
Epoch 00077: val_loss did not improve
3s - loss: 0.0039 - val_loss: 0.0037
Epoch 79/1500
Epoch 00078: val_loss did not improve
3s - loss: 0.0040 - val_loss: 0.0032
Epoch 80/1500
Epoch 00079: val_loss improved from 0.00279 to 0.00257, saving model to my_model.h5
3s - loss: 0.0037 - val_loss: 0.0026
Epoch 81/1500
Epoch 00080: val_loss improved from 0.00257 to 0.00227, saving model to my_model.h5
3s - loss: 0.0035 - val_loss: 0.0023
Epoch 82/1500
Epoch 00081: val_loss did not improve
3s - loss: 0.0035 - val_loss: 0.0025
Epoch 83/1500
Epoch 00082: val_loss improved from 0.00227 to 0.00223, saving model to my_model.h5
3s - loss: 0.0033 - val_loss: 0.0022
Epoch 84/1500
Epoch 00083: val_loss did not improve
3s - loss: 0.0033 - val_loss: 0.0023
Epoch 85/1500
Epoch 00084: val_loss improved from 0.00223 to 0.00220, saving model to my_model.h5
3s - loss: 0.0033 - val_loss: 0.0022
Epoch 86/1500
Epoch 00085: val_loss improved from 0.00220 to 0.00206, saving model to my_model.h5
3s - loss: 0.0033 - val_loss: 0.0021
Epoch 87/1500
Epoch 00086: val_loss improved from 0.00206 to 0.00201, saving model to my_model.h5
4s - loss: 0.0031 - val_loss: 0.0020
Epoch 88/1500
Epoch 00087: val_loss did not improve
3s - loss: 0.0031 - val_loss: 0.0022
Epoch 89/1500
Epoch 00088: val_loss did not improve
3s - loss: 0.0031 - val_loss: 0.0022
Epoch 90/1500
Epoch 00089: val_loss improved from 0.00201 to 0.00186, saving model to my_model.h5
3s - loss: 0.0030 - val_loss: 0.0019
Epoch 91/1500
Epoch 00090: val_loss did not improve
3s - loss: 0.0030 - val_loss: 0.0019
Epoch 92/1500
Epoch 00091: val_loss did not improve
3s - loss: 0.0030 - val_loss: 0.0019
Epoch 93/1500
Epoch 00092: val_loss improved from 0.00186 to 0.00182, saving model to my_model.h5
3s - loss: 0.0030 - val_loss: 0.0018
Epoch 94/1500
Epoch 00093: val_loss did not improve
3s - loss: 0.0030 - val_loss: 0.0020
Epoch 95/1500
Epoch 00094: val_loss did not improve
3s - loss: 0.0029 - val_loss: 0.0019
Epoch 96/1500
Epoch 00095: val_loss improved from 0.00182 to 0.00176, saving model to my_model.h5
4s - loss: 0.0028 - val_loss: 0.0018
Epoch 97/1500
Epoch 00096: val_loss improved from 0.00176 to 0.00167, saving model to my_model.h5
3s - loss: 0.0028 - val_loss: 0.0017
Epoch 98/1500
Epoch 00097: val_loss improved from 0.00167 to 0.00165, saving model to my_model.h5
3s - loss: 0.0029 - val_loss: 0.0016
Epoch 99/1500
Epoch 00098: val_loss did not improve
3s - loss: 0.0027 - val_loss: 0.0017
Epoch 100/1500
Epoch 00099: val_loss improved from 0.00165 to 0.00158, saving model to my_model.h5
3s - loss: 0.0028 - val_loss: 0.0016
Epoch 101/1500
Epoch 00100: val_loss did not improve
3s - loss: 0.0027 - val_loss: 0.0016
Epoch 102/1500
Epoch 00101: val_loss improved from 0.00158 to 0.00155, saving model to my_model.h5
3s - loss: 0.0027 - val_loss: 0.0015
Epoch 103/1500
Epoch 00102: val_loss did not improve
3s - loss: 0.0025 - val_loss: 0.0016
Epoch 104/1500
Epoch 00103: val_loss improved from 0.00155 to 0.00154, saving model to my_model.h5
3s - loss: 0.0027 - val_loss: 0.0015
Epoch 105/1500
Epoch 00104: val_loss did not improve
3s - loss: 0.0026 - val_loss: 0.0015
Epoch 106/1500
Epoch 00105: val_loss improved from 0.00154 to 0.00147, saving model to my_model.h5
3s - loss: 0.0026 - val_loss: 0.0015
Epoch 107/1500
Epoch 00106: val_loss did not improve
3s - loss: 0.0026 - val_loss: 0.0018
Epoch 108/1500
Epoch 00107: val_loss did not improve
3s - loss: 0.0026 - val_loss: 0.0015
Epoch 109/1500
Epoch 00108: val_loss improved from 0.00147 to 0.00140, saving model to my_model.h5
3s - loss: 0.0024 - val_loss: 0.0014
Epoch 110/1500
Epoch 00109: val_loss improved from 0.00140 to 0.00140, saving model to my_model.h5
3s - loss: 0.0024 - val_loss: 0.0014
Epoch 111/1500
Epoch 00110: val_loss improved from 0.00140 to 0.00138, saving model to my_model.h5
3s - loss: 0.0024 - val_loss: 0.0014
Epoch 112/1500
Epoch 00111: val_loss did not improve
3s - loss: 0.0024 - val_loss: 0.0015
Epoch 113/1500
Epoch 00112: val_loss did not improve
3s - loss: 0.0024 - val_loss: 0.0014
Epoch 114/1500
Epoch 00113: val_loss improved from 0.00138 to 0.00133, saving model to my_model.h5
3s - loss: 0.0024 - val_loss: 0.0013
Epoch 115/1500
Epoch 00114: val_loss improved from 0.00133 to 0.00132, saving model to my_model.h5
3s - loss: 0.0024 - val_loss: 0.0013
Epoch 116/1500
Epoch 00115: val_loss did not improve
3s - loss: 0.0023 - val_loss: 0.0014
Epoch 117/1500
Epoch 00116: val_loss improved from 0.00132 to 0.00132, saving model to my_model.h5
3s - loss: 0.0023 - val_loss: 0.0013
Epoch 118/1500
Epoch 00117: val_loss did not improve
3s - loss: 0.0024 - val_loss: 0.0014
Epoch 119/1500
Epoch 00118: val_loss did not improve
3s - loss: 0.0023 - val_loss: 0.0013
Epoch 120/1500
Epoch 00119: val_loss improved from 0.00132 to 0.00130, saving model to my_model.h5
3s - loss: 0.0022 - val_loss: 0.0013
Epoch 121/1500
Epoch 00120: val_loss improved from 0.00130 to 0.00128, saving model to my_model.h5
3s - loss: 0.0022 - val_loss: 0.0013
Epoch 122/1500
Epoch 00121: val_loss did not improve
3s - loss: 0.0022 - val_loss: 0.0013
Epoch 123/1500
Epoch 00122: val_loss did not improve
3s - loss: 0.0023 - val_loss: 0.0013
Epoch 124/1500
Epoch 00123: val_loss did not improve
3s - loss: 0.0022 - val_loss: 0.0013
Epoch 125/1500
Epoch 00124: val_loss did not improve
3s - loss: 0.0021 - val_loss: 0.0013
Epoch 126/1500
Epoch 00125: val_loss did not improve
3s - loss: 0.0023 - val_loss: 0.0014
Epoch 127/1500
Epoch 00126: val_loss improved from 0.00128 to 0.00124, saving model to my_model.h5
3s - loss: 0.0023 - val_loss: 0.0012
Epoch 128/1500
Epoch 00127: val_loss did not improve
3s - loss: 0.0022 - val_loss: 0.0014
Epoch 129/1500
Epoch 00128: val_loss improved from 0.00124 to 0.00122, saving model to my_model.h5
3s - loss: 0.0022 - val_loss: 0.0012
Epoch 130/1500
Epoch 00129: val_loss did not improve
3s - loss: 0.0022 - val_loss: 0.0012
Epoch 131/1500
Epoch 00130: val_loss improved from 0.00122 to 0.00117, saving model to my_model.h5
3s - loss: 0.0022 - val_loss: 0.0012
Epoch 132/1500
Epoch 00131: val_loss did not improve
3s - loss: 0.0021 - val_loss: 0.0012
Epoch 133/1500
Epoch 00132: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0012
Epoch 134/1500
Epoch 00133: val_loss did not improve
3s - loss: 0.0021 - val_loss: 0.0012
Epoch 135/1500
Epoch 00134: val_loss improved from 0.00117 to 0.00115, saving model to my_model.h5
3s - loss: 0.0021 - val_loss: 0.0011
Epoch 136/1500
Epoch 00135: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0012
Epoch 137/1500
Epoch 00136: val_loss improved from 0.00115 to 0.00114, saving model to my_model.h5
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 138/1500
Epoch 00137: val_loss improved from 0.00114 to 0.00113, saving model to my_model.h5
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 139/1500
Epoch 00138: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0013
Epoch 140/1500
Epoch 00139: val_loss improved from 0.00113 to 0.00112, saving model to my_model.h5
3s - loss: 0.0021 - val_loss: 0.0011
Epoch 141/1500
Epoch 00140: val_loss did not improve
3s - loss: 0.0021 - val_loss: 0.0012
Epoch 142/1500
Epoch 00141: val_loss improved from 0.00112 to 0.00112, saving model to my_model.h5
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 143/1500
Epoch 00142: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0012
Epoch 144/1500
Epoch 00143: val_loss improved from 0.00112 to 0.00109, saving model to my_model.h5
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 145/1500
Epoch 00144: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 146/1500
Epoch 00145: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 147/1500
Epoch 00146: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 148/1500
Epoch 00147: val_loss improved from 0.00109 to 0.00109, saving model to my_model.h5
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 149/1500
Epoch 00148: val_loss did not improve
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 150/1500
Epoch 00149: val_loss improved from 0.00109 to 0.00107, saving model to my_model.h5
3s - loss: 0.0020 - val_loss: 0.0011
Epoch 151/1500
Epoch 00150: val_loss improved from 0.00107 to 0.00107, saving model to my_model.h5
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 152/1500
Epoch 00151: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 153/1500
Epoch 00152: val_loss improved from 0.00107 to 0.00104, saving model to my_model.h5
6s - loss: 0.0019 - val_loss: 0.0010
Epoch 154/1500
Epoch 00153: val_loss did not improve
4s - loss: 0.0019 - val_loss: 0.0011
Epoch 155/1500
Epoch 00154: val_loss improved from 0.00104 to 0.00103, saving model to my_model.h5
3s - loss: 0.0019 - val_loss: 0.0010
Epoch 156/1500
Epoch 00155: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 157/1500
Epoch 00156: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0011
Epoch 158/1500
Epoch 00157: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0010
Epoch 159/1500
Epoch 00158: val_loss improved from 0.00103 to 0.00103, saving model to my_model.h5
3s - loss: 0.0018 - val_loss: 0.0010
Epoch 160/1500
Epoch 00159: val_loss improved from 0.00103 to 0.00102, saving model to my_model.h5
3s - loss: 0.0019 - val_loss: 0.0010
Epoch 161/1500
Epoch 00160: val_loss improved from 0.00102 to 0.00099, saving model to my_model.h5
3s - loss: 0.0018 - val_loss: 9.9125e-04
Epoch 162/1500
Epoch 00161: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0010
Epoch 163/1500
Epoch 00162: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0011
Epoch 164/1500
Epoch 00163: val_loss did not improve
3s - loss: 0.0019 - val_loss: 0.0010
Epoch 165/1500
Epoch 00164: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0011
Epoch 166/1500
Epoch 00165: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0011
Epoch 167/1500
Epoch 00166: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0010
Epoch 168/1500
Epoch 00167: val_loss improved from 0.00099 to 0.00096, saving model to my_model.h5
3s - loss: 0.0018 - val_loss: 9.5782e-04
Epoch 169/1500
Epoch 00168: val_loss did not improve
3s - loss: 0.0018 - val_loss: 9.9671e-04
Epoch 170/1500
Epoch 00169: val_loss did not improve
3s - loss: 0.0018 - val_loss: 0.0010
Epoch 171/1500
Epoch 00170: val_loss did not improve
3s - loss: 0.0018 - val_loss: 9.8167e-04
Epoch 172/1500
Epoch 00171: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.9253e-04
Epoch 173/1500
Epoch 00172: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.8815e-04
Epoch 174/1500
Epoch 00173: val_loss did not improve
3s - loss: 0.0018 - val_loss: 9.9718e-04
Epoch 175/1500
Epoch 00174: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.8071e-04
Epoch 176/1500
Epoch 00175: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.7539e-04
Epoch 177/1500
Epoch 00176: val_loss did not improve
3s - loss: 0.0017 - val_loss: 0.0010
Epoch 178/1500
Epoch 00177: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.8000e-04
Epoch 179/1500
Epoch 00178: val_loss improved from 0.00096 to 0.00096, saving model to my_model.h5
3s - loss: 0.0018 - val_loss: 9.5759e-04
Epoch 180/1500
Epoch 00179: val_loss improved from 0.00096 to 0.00092, saving model to my_model.h5
3s - loss: 0.0017 - val_loss: 9.2106e-04
Epoch 181/1500
Epoch 00180: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.6269e-04
Epoch 182/1500
Epoch 00181: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.3663e-04
Epoch 183/1500
Epoch 00182: val_loss improved from 0.00092 to 0.00092, saving model to my_model.h5
3s - loss: 0.0017 - val_loss: 9.1954e-04
Epoch 184/1500
Epoch 00183: val_loss improved from 0.00092 to 0.00090, saving model to my_model.h5
3s - loss: 0.0017 - val_loss: 8.9534e-04
Epoch 185/1500
Epoch 00184: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.7631e-04
Epoch 186/1500
Epoch 00185: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.7738e-04
Epoch 187/1500
Epoch 00186: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.0323e-04
Epoch 188/1500
Epoch 00187: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.7435e-04
Epoch 189/1500
Epoch 00188: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.7250e-04
Epoch 190/1500
Epoch 00189: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.1583e-04
Epoch 191/1500
Epoch 00190: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.0659e-04
Epoch 192/1500
Epoch 00191: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.2724e-04
Epoch 193/1500
Epoch 00192: val_loss did not improve
3s - loss: 0.0017 - val_loss: 9.3118e-04
Epoch 194/1500
Epoch 00193: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.5236e-04
Epoch 195/1500
Epoch 00194: val_loss improved from 0.00090 to 0.00089, saving model to my_model.h5
3s - loss: 0.0016 - val_loss: 8.9201e-04
Epoch 196/1500
Epoch 00195: val_loss did not improve
3s - loss: 0.0016 - val_loss: 0.0010
Epoch 197/1500
Epoch 00196: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.0330e-04
Epoch 198/1500
Epoch 00197: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.9365e-04
Epoch 199/1500
Epoch 00198: val_loss improved from 0.00089 to 0.00086, saving model to my_model.h5
3s - loss: 0.0016 - val_loss: 8.5934e-04
Epoch 200/1500
Epoch 00199: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.1249e-04
Epoch 201/1500
Epoch 00200: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.9316e-04
Epoch 202/1500
Epoch 00201: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.8503e-04
Epoch 203/1500
Epoch 00202: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.5988e-04
Epoch 204/1500
Epoch 00203: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.7863e-04
Epoch 205/1500
Epoch 00204: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.8603e-04
Epoch 206/1500
Epoch 00205: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.3276e-04
Epoch 207/1500
Epoch 00206: val_loss did not improve
3s - loss: 0.0016 - val_loss: 9.4261e-04
Epoch 208/1500
Epoch 00207: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.8836e-04
Epoch 209/1500
Epoch 00208: val_loss did not improve
3s - loss: 0.0015 - val_loss: 9.2911e-04
Epoch 210/1500
Epoch 00209: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.9785e-04
Epoch 211/1500
Epoch 00210: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.7665e-04
Epoch 212/1500
Epoch 00211: val_loss did not improve
3s - loss: 0.0015 - val_loss: 9.1245e-04
Epoch 213/1500
Epoch 00212: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.6512e-04
Epoch 214/1500
Epoch 00213: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.8770e-04
Epoch 215/1500
Epoch 00214: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.6712e-04
Epoch 216/1500
Epoch 00215: val_loss did not improve
3s - loss: 0.0016 - val_loss: 8.6932e-04
Epoch 217/1500
Epoch 00216: val_loss improved from 0.00086 to 0.00085, saving model to my_model.h5
3s - loss: 0.0015 - val_loss: 8.4515e-04
Epoch 218/1500
Epoch 00217: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.7688e-04
Epoch 219/1500
Epoch 00218: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.7378e-04
Epoch 220/1500
Epoch 00219: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.6192e-04
Epoch 221/1500
Epoch 00220: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.5098e-04
Epoch 222/1500
Epoch 00221: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.6529e-04
Epoch 223/1500
Epoch 00222: val_loss did not improve
3s - loss: 0.0015 - val_loss: 9.2543e-04
Epoch 224/1500
Epoch 00223: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.5053e-04
Epoch 225/1500
Epoch 00224: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.5800e-04
Epoch 226/1500
Epoch 00225: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.7354e-04
Epoch 227/1500
Epoch 00226: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.4899e-04
Epoch 228/1500
Epoch 00227: val_loss improved from 0.00085 to 0.00082, saving model to my_model.h5
3s - loss: 0.0015 - val_loss: 8.2073e-04
Epoch 229/1500
Epoch 00228: val_loss improved from 0.00082 to 0.00082, saving model to my_model.h5
3s - loss: 0.0015 - val_loss: 8.1711e-04
Epoch 230/1500
Epoch 00229: val_loss did not improve
3s - loss: 0.0014 - val_loss: 9.2542e-04
Epoch 231/1500
Epoch 00230: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.6518e-04
Epoch 232/1500
Epoch 00231: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.3898e-04
Epoch 233/1500
Epoch 00232: val_loss improved from 0.00082 to 0.00082, saving model to my_model.h5
3s - loss: 0.0015 - val_loss: 8.1608e-04
Epoch 234/1500
Epoch 00233: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.2205e-04
Epoch 235/1500
Epoch 00234: val_loss did not improve
3s - loss: 0.0015 - val_loss: 8.2513e-04
Epoch 236/1500
Epoch 00235: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.2612e-04
Epoch 237/1500
Epoch 00236: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.9325e-04
Epoch 238/1500
Epoch 00237: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.1799e-04
Epoch 239/1500
Epoch 00238: val_loss improved from 0.00082 to 0.00081, saving model to my_model.h5
3s - loss: 0.0014 - val_loss: 8.1220e-04
Epoch 240/1500
Epoch 00239: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.7636e-04
Epoch 241/1500
Epoch 00240: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.1379e-04
Epoch 242/1500
Epoch 00241: val_loss improved from 0.00081 to 0.00080, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.9996e-04
Epoch 243/1500
Epoch 00242: val_loss improved from 0.00080 to 0.00079, saving model to my_model.h5
3s - loss: 0.0014 - val_loss: 7.8680e-04
Epoch 244/1500
Epoch 00243: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.0873e-04
Epoch 245/1500
Epoch 00244: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.4088e-04
Epoch 246/1500
Epoch 00245: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.0362e-04
Epoch 247/1500
Epoch 00246: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.0135e-04
Epoch 248/1500
Epoch 00247: val_loss did not improve
3s - loss: 0.0014 - val_loss: 7.9996e-04
Epoch 249/1500
Epoch 00248: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.1285e-04
Epoch 250/1500
Epoch 00249: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.1171e-04
Epoch 251/1500
Epoch 00250: val_loss did not improve
3s - loss: 0.0014 - val_loss: 8.2047e-04
Epoch 252/1500
Epoch 00251: val_loss did not improve
3s - loss: 0.0014 - val_loss: 7.9724e-04
Epoch 253/1500
Epoch 00252: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.2711e-04
Epoch 254/1500
Epoch 00253: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.1367e-04
Epoch 255/1500
Epoch 00254: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.0454e-04
Epoch 256/1500
Epoch 00255: val_loss did not improve
3s - loss: 0.0014 - val_loss: 7.9288e-04
Epoch 257/1500
Epoch 00256: val_loss improved from 0.00079 to 0.00078, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.7711e-04
Epoch 258/1500
Epoch 00257: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8939e-04
Epoch 259/1500
Epoch 00258: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8926e-04
Epoch 260/1500
Epoch 00259: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8502e-04
Epoch 261/1500
Epoch 00260: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.7938e-04
Epoch 262/1500
Epoch 00261: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.2240e-04
Epoch 263/1500
Epoch 00262: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.1036e-04
Epoch 264/1500
Epoch 00263: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.9417e-04
Epoch 265/1500
Epoch 00264: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8514e-04
Epoch 266/1500
Epoch 00265: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8713e-04
Epoch 267/1500
Epoch 00266: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.0368e-04
Epoch 268/1500
Epoch 00267: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8991e-04
Epoch 269/1500
Epoch 00268: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.7711e-04
Epoch 270/1500
Epoch 00269: val_loss did not improve
3s - loss: 0.0013 - val_loss: 8.0170e-04
Epoch 271/1500
Epoch 00270: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.9354e-04
Epoch 272/1500
Epoch 00271: val_loss improved from 0.00078 to 0.00077, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.6841e-04
Epoch 273/1500
Epoch 00272: val_loss improved from 0.00077 to 0.00074, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.4360e-04
Epoch 274/1500
Epoch 00273: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8408e-04
Epoch 275/1500
Epoch 00274: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.8468e-04
Epoch 276/1500
Epoch 00275: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.6492e-04
Epoch 277/1500
Epoch 00276: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.5283e-04
Epoch 278/1500
Epoch 00277: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.5164e-04
Epoch 279/1500
Epoch 00278: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.4984e-04
Epoch 280/1500
Epoch 00279: val_loss improved from 0.00074 to 0.00074, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.3961e-04
Epoch 281/1500
Epoch 00280: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.5201e-04
Epoch 282/1500
Epoch 00281: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4652e-04
Epoch 283/1500
Epoch 00282: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.5559e-04
Epoch 284/1500
Epoch 00283: val_loss improved from 0.00074 to 0.00074, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.3905e-04
Epoch 285/1500
Epoch 00284: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.6042e-04
Epoch 286/1500
Epoch 00285: val_loss improved from 0.00074 to 0.00074, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.3546e-04
Epoch 287/1500
Epoch 00286: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.4781e-04
Epoch 288/1500
Epoch 00287: val_loss improved from 0.00074 to 0.00074, saving model to my_model.h5
3s - loss: 0.0013 - val_loss: 7.3545e-04
Epoch 289/1500
Epoch 00288: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.5916e-04
Epoch 290/1500
Epoch 00289: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.7142e-04
Epoch 291/1500
Epoch 00290: val_loss improved from 0.00074 to 0.00073, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.3277e-04
Epoch 292/1500
Epoch 00291: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4579e-04
Epoch 293/1500
Epoch 00292: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.3357e-04
Epoch 294/1500
Epoch 00293: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.6056e-04
Epoch 295/1500
Epoch 00294: val_loss did not improve
3s - loss: 0.0013 - val_loss: 7.3551e-04
Epoch 296/1500
Epoch 00295: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.3517e-04
Epoch 297/1500
Epoch 00296: val_loss improved from 0.00073 to 0.00073, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.2884e-04
Epoch 298/1500
Epoch 00297: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.9437e-04
Epoch 299/1500
Epoch 00298: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.3397e-04
Epoch 300/1500
Epoch 00299: val_loss improved from 0.00073 to 0.00073, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.2796e-04
Epoch 301/1500
Epoch 00300: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.3100e-04
Epoch 302/1500
Epoch 00301: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4192e-04
Epoch 303/1500
Epoch 00302: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4415e-04
Epoch 304/1500
Epoch 00303: val_loss improved from 0.00073 to 0.00072, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.1713e-04
Epoch 305/1500
Epoch 00304: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4028e-04
Epoch 306/1500
Epoch 00305: val_loss improved from 0.00072 to 0.00072, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.1581e-04
Epoch 307/1500
Epoch 00306: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.7058e-04
Epoch 308/1500
Epoch 00307: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.2563e-04
Epoch 309/1500
Epoch 00308: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.2485e-04
Epoch 310/1500
Epoch 00309: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.2453e-04
Epoch 311/1500
Epoch 00310: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.4046e-04
Epoch 312/1500
Epoch 00311: val_loss improved from 0.00072 to 0.00072, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 7.1575e-04
Epoch 313/1500
Epoch 00312: val_loss improved from 0.00072 to 0.00070, saving model to my_model.h5
3s - loss: 0.0012 - val_loss: 6.9679e-04
Epoch 314/1500
Epoch 00313: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0371e-04
Epoch 315/1500
Epoch 00314: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.3030e-04
Epoch 316/1500
Epoch 00315: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.5382e-04
Epoch 317/1500
Epoch 00316: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1519e-04
Epoch 318/1500
Epoch 00317: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.2914e-04
Epoch 319/1500
Epoch 00318: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1272e-04
Epoch 320/1500
Epoch 00319: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0794e-04
Epoch 321/1500
Epoch 00320: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1550e-04
Epoch 322/1500
Epoch 00321: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1621e-04
Epoch 323/1500
Epoch 00322: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.2063e-04
Epoch 324/1500
Epoch 00323: val_loss improved from 0.00070 to 0.00069, saving model to my_model.h5
3s - loss: 0.0011 - val_loss: 6.9449e-04
Epoch 325/1500
Epoch 00324: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1252e-04
Epoch 326/1500
Epoch 00325: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0842e-04
Epoch 327/1500
Epoch 00326: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0292e-04
Epoch 328/1500
Epoch 00327: val_loss improved from 0.00069 to 0.00068, saving model to my_model.h5
3s - loss: 0.0011 - val_loss: 6.8439e-04
Epoch 329/1500
Epoch 00328: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0757e-04
Epoch 330/1500
Epoch 00329: val_loss did not improve
3s - loss: 0.0012 - val_loss: 7.1285e-04
Epoch 331/1500
Epoch 00330: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0321e-04
Epoch 332/1500
Epoch 00331: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.9534e-04
Epoch 333/1500
Epoch 00332: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0241e-04
Epoch 334/1500
Epoch 00333: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.2159e-04
Epoch 335/1500
Epoch 00334: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0328e-04
Epoch 336/1500
Epoch 00335: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1074e-04
Epoch 337/1500
Epoch 00336: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.9043e-04
Epoch 338/1500
Epoch 00337: val_loss improved from 0.00068 to 0.00067, saving model to my_model.h5
3s - loss: 0.0011 - val_loss: 6.7441e-04
Epoch 339/1500
Epoch 00338: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.7704e-04
Epoch 340/1500
Epoch 00339: val_loss did not improve
3s - loss: 0.0010 - val_loss: 7.1044e-04
Epoch 341/1500
Epoch 00340: val_loss improved from 0.00067 to 0.00067, saving model to my_model.h5
3s - loss: 0.0011 - val_loss: 6.7224e-04
Epoch 342/1500
Epoch 00341: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.1225e-04
Epoch 343/1500
Epoch 00342: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.7252e-04
Epoch 344/1500
Epoch 00343: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.8861e-04
Epoch 345/1500
Epoch 00344: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0348e-04
Epoch 346/1500
Epoch 00345: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.9397e-04
Epoch 347/1500
Epoch 00346: val_loss did not improve
3s - loss: 0.0011 - val_loss: 7.0296e-04
Epoch 348/1500
Epoch 00347: val_loss did not improve
3s - loss: 0.0010 - val_loss: 7.4490e-04
Epoch 349/1500
Epoch 00348: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.7573e-04
Epoch 350/1500
Epoch 00349: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.8280e-04
Epoch 351/1500
Epoch 00350: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.8534e-04
Epoch 352/1500
Epoch 00351: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.8028e-04
Epoch 353/1500
Epoch 00352: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.9112e-04
Epoch 354/1500
Epoch 00353: val_loss did not improve
3s - loss: 0.0011 - val_loss: 6.7840e-04
Epoch 355/1500
Epoch 00354: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.9388e-04
Epoch 356/1500
Epoch 00355: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.8581e-04
Epoch 357/1500
Epoch 00356: val_loss improved from 0.00067 to 0.00067, saving model to my_model.h5
3s - loss: 0.0010 - val_loss: 6.7177e-04
Epoch 358/1500
Epoch 00357: val_loss improved from 0.00067 to 0.00066, saving model to my_model.h5
3s - loss: 0.0011 - val_loss: 6.6323e-04
Epoch 359/1500
Epoch 00358: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.7382e-04
Epoch 360/1500
Epoch 00359: val_loss did not improve
3s - loss: 0.0010 - val_loss: 7.0716e-04
Epoch 361/1500
Epoch 00360: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.9655e-04
Epoch 362/1500
Epoch 00361: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.9725e-04
Epoch 363/1500
Epoch 00362: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.6858e-04
Epoch 364/1500
Epoch 00363: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.7364e-04
Epoch 365/1500
Epoch 00364: val_loss improved from 0.00066 to 0.00066, saving model to my_model.h5
3s - loss: 0.0010 - val_loss: 6.6304e-04
Epoch 366/1500
Epoch 00365: val_loss did not improve
3s - loss: 9.7641e-04 - val_loss: 6.6427e-04
Epoch 367/1500
Epoch 00366: val_loss did not improve
3s - loss: 9.7060e-04 - val_loss: 7.0062e-04
Epoch 368/1500
Epoch 00367: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.8613e-04
Epoch 369/1500
Epoch 00368: val_loss did not improve
3s - loss: 9.9906e-04 - val_loss: 6.9158e-04
Epoch 370/1500
Epoch 00369: val_loss did not improve
3s - loss: 9.7882e-04 - val_loss: 6.6359e-04
Epoch 371/1500
Epoch 00370: val_loss improved from 0.00066 to 0.00065, saving model to my_model.h5
3s - loss: 0.0010 - val_loss: 6.5480e-04
Epoch 372/1500
Epoch 00371: val_loss did not improve
3s - loss: 9.9083e-04 - val_loss: 6.6186e-04
Epoch 373/1500
Epoch 00372: val_loss did not improve
3s - loss: 9.9927e-04 - val_loss: 6.5510e-04
Epoch 374/1500
Epoch 00373: val_loss did not improve
3s - loss: 9.9001e-04 - val_loss: 6.7200e-04
Epoch 375/1500
Epoch 00374: val_loss did not improve
3s - loss: 9.6387e-04 - val_loss: 6.8273e-04
Epoch 376/1500
Epoch 00375: val_loss did not improve
3s - loss: 9.9645e-04 - val_loss: 6.8326e-04
Epoch 377/1500
Epoch 00376: val_loss did not improve
3s - loss: 9.9917e-04 - val_loss: 6.7650e-04
Epoch 378/1500
Epoch 00377: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.6325e-04
Epoch 379/1500
Epoch 00378: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.6317e-04
Epoch 380/1500
Epoch 00379: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.7509e-04
Epoch 381/1500
Epoch 00380: val_loss did not improve
3s - loss: 9.8707e-04 - val_loss: 6.6310e-04
Epoch 382/1500
Epoch 00381: val_loss did not improve
3s - loss: 9.5767e-04 - val_loss: 6.7001e-04
Epoch 383/1500
Epoch 00382: val_loss did not improve
3s - loss: 9.8763e-04 - val_loss: 6.5700e-04
Epoch 384/1500
Epoch 00383: val_loss did not improve
3s - loss: 9.7531e-04 - val_loss: 6.7956e-04
Epoch 385/1500
Epoch 00384: val_loss improved from 0.00065 to 0.00065, saving model to my_model.h5
3s - loss: 9.5182e-04 - val_loss: 6.5219e-04
Epoch 386/1500
Epoch 00385: val_loss did not improve
3s - loss: 9.7138e-04 - val_loss: 6.6270e-04
Epoch 387/1500
Epoch 00386: val_loss did not improve
3s - loss: 9.5890e-04 - val_loss: 6.6193e-04
Epoch 388/1500
Epoch 00387: val_loss did not improve
3s - loss: 0.0010 - val_loss: 6.5303e-04
Epoch 389/1500
Epoch 00388: val_loss did not improve
3s - loss: 9.5593e-04 - val_loss: 6.6237e-04
Epoch 390/1500
Epoch 00389: val_loss did not improve
3s - loss: 9.6215e-04 - val_loss: 6.5585e-04
Epoch 391/1500
Epoch 00390: val_loss did not improve
3s - loss: 9.0396e-04 - val_loss: 6.6676e-04
Epoch 392/1500
Epoch 00391: val_loss improved from 0.00065 to 0.00064, saving model to my_model.h5
3s - loss: 9.5869e-04 - val_loss: 6.4273e-04
Epoch 393/1500
Epoch 00392: val_loss did not improve
3s - loss: 9.7143e-04 - val_loss: 6.5103e-04
Epoch 394/1500
Epoch 00393: val_loss did not improve
3s - loss: 9.3385e-04 - val_loss: 6.7118e-04
Epoch 395/1500
Epoch 00394: val_loss did not improve
3s - loss: 9.9011e-04 - val_loss: 6.6054e-04
Epoch 396/1500
Epoch 00395: val_loss did not improve
3s - loss: 9.8537e-04 - val_loss: 6.8153e-04
Epoch 397/1500
Epoch 00396: val_loss did not improve
3s - loss: 9.8043e-04 - val_loss: 6.5719e-04
Epoch 398/1500
Epoch 00397: val_loss did not improve
3s - loss: 9.2762e-04 - val_loss: 6.6988e-04
Epoch 399/1500
Epoch 00398: val_loss did not improve
3s - loss: 9.3276e-04 - val_loss: 6.5932e-04
Epoch 400/1500
Epoch 00399: val_loss did not improve
3s - loss: 9.2507e-04 - val_loss: 6.6358e-04
Epoch 401/1500
Epoch 00400: val_loss improved from 0.00064 to 0.00063, saving model to my_model.h5
3s - loss: 9.6304e-04 - val_loss: 6.2574e-04
Epoch 402/1500
Epoch 00401: val_loss improved from 0.00063 to 0.00062, saving model to my_model.h5
3s - loss: 9.4564e-04 - val_loss: 6.2156e-04
Epoch 403/1500
Epoch 00402: val_loss did not improve
3s - loss: 9.1425e-04 - val_loss: 6.6629e-04
Epoch 404/1500
Epoch 00403: val_loss did not improve
3s - loss: 9.0342e-04 - val_loss: 6.6656e-04
Epoch 405/1500
Epoch 00404: val_loss did not improve
3s - loss: 9.2420e-04 - val_loss: 6.5062e-04
Epoch 406/1500
Epoch 00405: val_loss did not improve
3s - loss: 9.2706e-04 - val_loss: 6.3398e-04
Epoch 407/1500
Epoch 00406: val_loss did not improve
3s - loss: 9.4370e-04 - val_loss: 6.5127e-04
Epoch 408/1500
Epoch 00407: val_loss did not improve
3s - loss: 8.8484e-04 - val_loss: 6.4088e-04
Epoch 409/1500
Epoch 00408: val_loss did not improve
3s - loss: 9.2530e-04 - val_loss: 6.4464e-04
Epoch 410/1500
Epoch 00409: val_loss did not improve
3s - loss: 9.1436e-04 - val_loss: 6.5008e-04
Epoch 411/1500
Epoch 00410: val_loss did not improve
3s - loss: 9.1637e-04 - val_loss: 6.8116e-04
Epoch 412/1500
Epoch 00411: val_loss did not improve
3s - loss: 8.9990e-04 - val_loss: 6.3605e-04
Epoch 413/1500
Epoch 00412: val_loss did not improve
3s - loss: 9.4456e-04 - val_loss: 6.5252e-04
Epoch 414/1500
Epoch 00413: val_loss did not improve
3s - loss: 8.9412e-04 - val_loss: 6.3551e-04
Epoch 415/1500
Epoch 00414: val_loss did not improve
3s - loss: 9.3806e-04 - val_loss: 6.6050e-04
Epoch 416/1500
Epoch 00415: val_loss did not improve
3s - loss: 8.9991e-04 - val_loss: 6.5259e-04
Epoch 417/1500
Epoch 00416: val_loss did not improve
3s - loss: 8.8472e-04 - val_loss: 6.3522e-04
Epoch 418/1500
Epoch 00417: val_loss did not improve
3s - loss: 8.9132e-04 - val_loss: 6.4105e-04
Epoch 419/1500
Epoch 00418: val_loss did not improve
3s - loss: 8.9965e-04 - val_loss: 6.5162e-04
Epoch 420/1500
Epoch 00419: val_loss did not improve
3s - loss: 9.0048e-04 - val_loss: 6.7598e-04
Epoch 421/1500
Epoch 00420: val_loss did not improve
3s - loss: 9.1802e-04 - val_loss: 6.4603e-04
Epoch 422/1500
Epoch 00421: val_loss did not improve
3s - loss: 8.6420e-04 - val_loss: 6.2796e-04
Epoch 423/1500
Epoch 00422: val_loss did not improve
3s - loss: 8.4032e-04 - val_loss: 6.5623e-04
Epoch 424/1500
Epoch 00423: val_loss did not improve
3s - loss: 9.0778e-04 - val_loss: 6.5806e-04
Epoch 425/1500
Epoch 00424: val_loss did not improve
3s - loss: 9.4154e-04 - val_loss: 6.3663e-04
Epoch 426/1500
Epoch 00425: val_loss did not improve
3s - loss: 9.2535e-04 - val_loss: 6.4083e-04
Epoch 427/1500
Epoch 00426: val_loss improved from 0.00062 to 0.00062, saving model to my_model.h5
3s - loss: 8.9775e-04 - val_loss: 6.2024e-04
Epoch 428/1500
Epoch 00427: val_loss did not improve
3s - loss: 8.8755e-04 - val_loss: 6.2821e-04
Epoch 429/1500
Epoch 00428: val_loss did not improve
3s - loss: 8.8274e-04 - val_loss: 6.3495e-04
Epoch 430/1500
Epoch 00429: val_loss did not improve
3s - loss: 8.9077e-04 - val_loss: 6.5225e-04
Epoch 431/1500
Epoch 00430: val_loss did not improve
3s - loss: 8.3390e-04 - val_loss: 6.4332e-04
Epoch 432/1500
Epoch 00431: val_loss did not improve
3s - loss: 8.6196e-04 - val_loss: 6.3569e-04
Epoch 433/1500
Epoch 00432: val_loss did not improve
3s - loss: 8.4448e-04 - val_loss: 6.5886e-04
Epoch 434/1500
Epoch 00433: val_loss did not improve
3s - loss: 8.8392e-04 - val_loss: 6.7427e-04
Epoch 435/1500
Epoch 00434: val_loss did not improve
3s - loss: 8.7719e-04 - val_loss: 6.3190e-04
Epoch 436/1500
Epoch 00435: val_loss did not improve
3s - loss: 8.6976e-04 - val_loss: 6.2052e-04
Epoch 437/1500
Epoch 00436: val_loss did not improve
3s - loss: 8.2719e-04 - val_loss: 6.4249e-04
Epoch 438/1500
Epoch 00437: val_loss did not improve
3s - loss: 8.8126e-04 - val_loss: 6.3926e-04
Epoch 439/1500
Epoch 00438: val_loss did not improve
3s - loss: 8.5641e-04 - val_loss: 6.2836e-04
Epoch 440/1500
Epoch 00439: val_loss did not improve
3s - loss: 8.5360e-04 - val_loss: 6.5204e-04
Epoch 441/1500
Epoch 00440: val_loss did not improve
3s - loss: 8.7210e-04 - val_loss: 6.4637e-04
Epoch 442/1500
Epoch 00441: val_loss improved from 0.00062 to 0.00062, saving model to my_model.h5
3s - loss: 8.7590e-04 - val_loss: 6.1964e-04
Epoch 443/1500
Epoch 00442: val_loss improved from 0.00062 to 0.00061, saving model to my_model.h5
3s - loss: 8.6415e-04 - val_loss: 6.1463e-04
Epoch 444/1500
Epoch 00443: val_loss did not improve
3s - loss: 8.8720e-04 - val_loss: 6.3212e-04
Epoch 445/1500
Epoch 00444: val_loss did not improve
3s - loss: 8.5161e-04 - val_loss: 6.3428e-04
Epoch 446/1500
Epoch 00445: val_loss did not improve
3s - loss: 8.6548e-04 - val_loss: 6.1597e-04
Epoch 447/1500
Epoch 00446: val_loss did not improve
3s - loss: 8.4815e-04 - val_loss: 6.3083e-04
Epoch 448/1500
Epoch 00447: val_loss did not improve
3s - loss: 8.4861e-04 - val_loss: 6.6419e-04
Epoch 449/1500
Epoch 00448: val_loss did not improve
3s - loss: 8.4314e-04 - val_loss: 6.2052e-04
Epoch 450/1500
Epoch 00449: val_loss did not improve
3s - loss: 8.4036e-04 - val_loss: 6.1833e-04
Epoch 451/1500
Epoch 00450: val_loss did not improve
3s - loss: 8.6658e-04 - val_loss: 6.1695e-04
Epoch 452/1500
Epoch 00451: val_loss did not improve
3s - loss: 8.2771e-04 - val_loss: 6.2399e-04
Epoch 453/1500
Epoch 00452: val_loss did not improve
3s - loss: 8.4613e-04 - val_loss: 6.1980e-04
Epoch 454/1500
Epoch 00453: val_loss did not improve
3s - loss: 8.3147e-04 - val_loss: 6.3408e-04
Epoch 455/1500
Epoch 00454: val_loss did not improve
3s - loss: 8.5037e-04 - val_loss: 6.2611e-04
Epoch 456/1500
Epoch 00455: val_loss did not improve
3s - loss: 8.4203e-04 - val_loss: 6.1591e-04
Epoch 457/1500
Epoch 00456: val_loss did not improve
3s - loss: 8.5589e-04 - val_loss: 6.2132e-04
Epoch 458/1500
Epoch 00457: val_loss did not improve
3s - loss: 8.6766e-04 - val_loss: 6.5520e-04
Epoch 459/1500
Epoch 00458: val_loss did not improve
3s - loss: 8.4131e-04 - val_loss: 6.3138e-04
Epoch 460/1500
Epoch 00459: val_loss improved from 0.00061 to 0.00061, saving model to my_model.h5
3s - loss: 8.5627e-04 - val_loss: 6.1091e-04
Epoch 461/1500
Epoch 00460: val_loss did not improve
3s - loss: 8.4384e-04 - val_loss: 6.3297e-04
Epoch 462/1500
Epoch 00461: val_loss did not improve
3s - loss: 8.4208e-04 - val_loss: 6.4250e-04
Epoch 463/1500
Epoch 00462: val_loss did not improve
3s - loss: 7.9607e-04 - val_loss: 6.3353e-04
Epoch 464/1500
Epoch 00463: val_loss did not improve
3s - loss: 8.1977e-04 - val_loss: 6.2115e-04
Epoch 465/1500
Epoch 00464: val_loss did not improve
3s - loss: 8.3242e-04 - val_loss: 6.1635e-04
Epoch 466/1500
Epoch 00465: val_loss did not improve
3s - loss: 8.3122e-04 - val_loss: 6.2974e-04
Epoch 467/1500
Epoch 00466: val_loss did not improve
3s - loss: 8.1340e-04 - val_loss: 6.1475e-04
Epoch 468/1500
Epoch 00467: val_loss improved from 0.00061 to 0.00061, saving model to my_model.h5
3s - loss: 8.2285e-04 - val_loss: 6.0534e-04
Epoch 469/1500
Epoch 00468: val_loss did not improve
3s - loss: 8.2799e-04 - val_loss: 6.1179e-04
Epoch 470/1500
Epoch 00469: val_loss did not improve
3s - loss: 8.3732e-04 - val_loss: 6.0748e-04
Epoch 471/1500
Epoch 00470: val_loss did not improve
3s - loss: 8.4346e-04 - val_loss: 6.4820e-04
Epoch 472/1500
Epoch 00471: val_loss did not improve
3s - loss: 8.1343e-04 - val_loss: 6.3752e-04
Epoch 473/1500
Epoch 00472: val_loss did not improve
3s - loss: 8.1961e-04 - val_loss: 6.1485e-04
Epoch 474/1500
Epoch 00473: val_loss did not improve
3s - loss: 8.3178e-04 - val_loss: 6.2266e-04
Epoch 475/1500
Epoch 00474: val_loss did not improve
3s - loss: 8.0911e-04 - val_loss: 6.1715e-04
Epoch 476/1500
Epoch 00475: val_loss improved from 0.00061 to 0.00060, saving model to my_model.h5
3s - loss: 8.6821e-04 - val_loss: 5.9808e-04
Epoch 477/1500
Epoch 00476: val_loss did not improve
3s - loss: 8.3288e-04 - val_loss: 6.0967e-04
Epoch 478/1500
Epoch 00477: val_loss did not improve
3s - loss: 7.7857e-04 - val_loss: 6.0651e-04
Epoch 479/1500
Epoch 00478: val_loss did not improve
3s - loss: 7.7257e-04 - val_loss: 6.4658e-04
Epoch 480/1500
Epoch 00479: val_loss did not improve
3s - loss: 8.3324e-04 - val_loss: 6.2087e-04
Epoch 481/1500
Epoch 00480: val_loss did not improve
3s - loss: 8.4103e-04 - val_loss: 6.4169e-04
Epoch 482/1500
Epoch 00481: val_loss did not improve
3s - loss: 7.9220e-04 - val_loss: 6.1217e-04
Epoch 483/1500
Epoch 00482: val_loss did not improve
3s - loss: 8.1378e-04 - val_loss: 6.0945e-04
Epoch 484/1500
Epoch 00483: val_loss did not improve
3s - loss: 8.1084e-04 - val_loss: 6.2325e-04
Epoch 485/1500
Epoch 00484: val_loss did not improve
3s - loss: 8.3256e-04 - val_loss: 6.3859e-04
Epoch 486/1500
Epoch 00485: val_loss did not improve
3s - loss: 8.0165e-04 - val_loss: 6.4101e-04
Epoch 487/1500
Epoch 00486: val_loss did not improve
3s - loss: 7.8870e-04 - val_loss: 6.2655e-04
Epoch 488/1500
Epoch 00487: val_loss did not improve
3s - loss: 8.1837e-04 - val_loss: 6.1318e-04
Epoch 489/1500
Epoch 00488: val_loss did not improve
3s - loss: 8.1955e-04 - val_loss: 6.1362e-04
Epoch 490/1500
Epoch 00489: val_loss did not improve
3s - loss: 7.7252e-04 - val_loss: 6.1913e-04
Epoch 491/1500
Epoch 00490: val_loss did not improve
3s - loss: 8.1365e-04 - val_loss: 6.2196e-04
Epoch 492/1500
Epoch 00491: val_loss did not improve
3s - loss: 7.9228e-04 - val_loss: 6.1065e-04
Epoch 493/1500
Epoch 00492: val_loss did not improve
3s - loss: 7.8067e-04 - val_loss: 6.2462e-04
Epoch 494/1500
Epoch 00493: val_loss did not improve
3s - loss: 8.1638e-04 - val_loss: 6.0256e-04
Epoch 495/1500
Epoch 00494: val_loss did not improve
3s - loss: 8.3617e-04 - val_loss: 6.0803e-04
Epoch 496/1500
Epoch 00495: val_loss did not improve
3s - loss: 8.0330e-04 - val_loss: 5.9905e-04
Epoch 497/1500
Epoch 00496: val_loss did not improve
3s - loss: 8.0468e-04 - val_loss: 6.0693e-04
Epoch 498/1500
Epoch 00497: val_loss improved from 0.00060 to 0.00059, saving model to my_model.h5
3s - loss: 7.9681e-04 - val_loss: 5.9125e-04
Epoch 499/1500
Epoch 00498: val_loss did not improve
3s - loss: 7.7771e-04 - val_loss: 6.1363e-04
Epoch 500/1500
Epoch 00499: val_loss did not improve
3s - loss: 8.0245e-04 - val_loss: 6.2731e-04
Epoch 501/1500
Epoch 00500: val_loss did not improve
3s - loss: 7.9963e-04 - val_loss: 5.9504e-04
Epoch 502/1500
Epoch 00501: val_loss did not improve
3s - loss: 7.7636e-04 - val_loss: 6.0602e-04
Epoch 503/1500
Epoch 00502: val_loss did not improve
3s - loss: 7.5338e-04 - val_loss: 6.1057e-04
Epoch 504/1500
Epoch 00503: val_loss did not improve
3s - loss: 7.7348e-04 - val_loss: 6.0286e-04
Epoch 505/1500
Epoch 00504: val_loss did not improve
3s - loss: 7.9684e-04 - val_loss: 5.9988e-04
Epoch 506/1500
Epoch 00505: val_loss did not improve
3s - loss: 7.8635e-04 - val_loss: 6.1504e-04
Epoch 507/1500
Epoch 00506: val_loss did not improve
3s - loss: 7.7731e-04 - val_loss: 6.0203e-04
Epoch 508/1500
Epoch 00507: val_loss improved from 0.00059 to 0.00059, saving model to my_model.h5
3s - loss: 8.2347e-04 - val_loss: 5.8585e-04
Epoch 509/1500
Epoch 00508: val_loss improved from 0.00059 to 0.00058, saving model to my_model.h5
3s - loss: 7.9683e-04 - val_loss: 5.8219e-04
Epoch 510/1500
Epoch 00509: val_loss did not improve
3s - loss: 7.7186e-04 - val_loss: 5.9754e-04
Epoch 511/1500
Epoch 00510: val_loss did not improve
3s - loss: 7.6426e-04 - val_loss: 6.0848e-04
Epoch 512/1500
Epoch 00511: val_loss did not improve
3s - loss: 7.7609e-04 - val_loss: 5.9423e-04
Epoch 513/1500
Epoch 00512: val_loss did not improve
3s - loss: 7.5563e-04 - val_loss: 6.1947e-04
Epoch 514/1500
Epoch 00513: val_loss did not improve
3s - loss: 7.6937e-04 - val_loss: 5.9168e-04
Epoch 515/1500
Epoch 00514: val_loss did not improve
3s - loss: 7.8078e-04 - val_loss: 6.0456e-04
Epoch 516/1500
Epoch 00515: val_loss did not improve
3s - loss: 8.0217e-04 - val_loss: 5.9245e-04
Epoch 517/1500
Epoch 00516: val_loss did not improve
3s - loss: 7.6168e-04 - val_loss: 6.0182e-04
Epoch 518/1500
Epoch 00517: val_loss did not improve
3s - loss: 7.4794e-04 - val_loss: 5.9074e-04
Epoch 519/1500
Epoch 00518: val_loss did not improve
3s - loss: 7.6133e-04 - val_loss: 5.9982e-04
Epoch 520/1500
Epoch 00519: val_loss did not improve
3s - loss: 7.5231e-04 - val_loss: 5.9423e-04
Epoch 521/1500
Epoch 00520: val_loss did not improve
3s - loss: 7.8972e-04 - val_loss: 6.1909e-04
Epoch 522/1500
Epoch 00521: val_loss did not improve
3s - loss: 7.9527e-04 - val_loss: 5.8424e-04
Epoch 523/1500
Epoch 00522: val_loss did not improve
3s - loss: 8.2745e-04 - val_loss: 5.8648e-04
Epoch 524/1500
Epoch 00523: val_loss did not improve
3s - loss: 7.7338e-04 - val_loss: 5.9959e-04
Epoch 525/1500
Epoch 00524: val_loss did not improve
3s - loss: 7.5753e-04 - val_loss: 6.0903e-04
Epoch 526/1500
Epoch 00525: val_loss did not improve
3s - loss: 7.6474e-04 - val_loss: 6.1902e-04
Epoch 527/1500
Epoch 00526: val_loss did not improve
3s - loss: 7.6177e-04 - val_loss: 6.2921e-04
Epoch 528/1500
Epoch 00527: val_loss did not improve
3s - loss: 7.4986e-04 - val_loss: 6.0927e-04
Epoch 529/1500
Epoch 00528: val_loss did not improve
3s - loss: 7.9434e-04 - val_loss: 6.0727e-04
Epoch 530/1500
Epoch 00529: val_loss did not improve
3s - loss: 7.5949e-04 - val_loss: 6.0594e-04
Epoch 531/1500
Epoch 00530: val_loss did not improve
3s - loss: 7.7305e-04 - val_loss: 5.8577e-04
Epoch 532/1500
Epoch 00531: val_loss did not improve
3s - loss: 7.6767e-04 - val_loss: 5.8926e-04
Epoch 533/1500
Epoch 00532: val_loss did not improve
3s - loss: 7.8541e-04 - val_loss: 5.9223e-04
Epoch 534/1500
Epoch 00533: val_loss did not improve
3s - loss: 7.3499e-04 - val_loss: 6.1448e-04
Epoch 535/1500
Epoch 00534: val_loss did not improve
3s - loss: 7.6889e-04 - val_loss: 6.1763e-04
Epoch 536/1500
Epoch 00535: val_loss did not improve
3s - loss: 7.6282e-04 - val_loss: 5.9470e-04
Epoch 537/1500
Epoch 00536: val_loss did not improve
3s - loss: 7.0723e-04 - val_loss: 5.8530e-04
Epoch 538/1500
Epoch 00537: val_loss did not improve
3s - loss: 7.7192e-04 - val_loss: 5.9444e-04
Epoch 539/1500
Epoch 00538: val_loss did not improve
3s - loss: 7.5943e-04 - val_loss: 5.8553e-04
Epoch 540/1500
Epoch 00539: val_loss improved from 0.00058 to 0.00058, saving model to my_model.h5
3s - loss: 7.4163e-04 - val_loss: 5.7509e-04
Epoch 541/1500
Epoch 00540: val_loss did not improve
3s - loss: 7.5266e-04 - val_loss: 5.9938e-04
Epoch 542/1500
Epoch 00541: val_loss did not improve
3s - loss: 7.3620e-04 - val_loss: 6.0315e-04
Epoch 543/1500
Epoch 00542: val_loss did not improve
3s - loss: 7.4908e-04 - val_loss: 6.0084e-04
Epoch 544/1500
Epoch 00543: val_loss did not improve
3s - loss: 7.5829e-04 - val_loss: 5.9394e-04
Epoch 545/1500
Epoch 00544: val_loss did not improve
3s - loss: 7.2358e-04 - val_loss: 6.0356e-04
Epoch 546/1500
Epoch 00545: val_loss did not improve
3s - loss: 7.6915e-04 - val_loss: 5.9458e-04
Epoch 547/1500
Epoch 00546: val_loss did not improve
3s - loss: 7.3712e-04 - val_loss: 5.8560e-04
Epoch 548/1500
Epoch 00547: val_loss did not improve
3s - loss: 7.1662e-04 - val_loss: 5.9379e-04
Epoch 549/1500
Epoch 00548: val_loss did not improve
3s - loss: 7.3606e-04 - val_loss: 5.9577e-04
Epoch 550/1500
Epoch 00549: val_loss did not improve
3s - loss: 7.5722e-04 - val_loss: 5.9393e-04
Epoch 551/1500
Epoch 00550: val_loss did not improve
3s - loss: 7.5995e-04 - val_loss: 5.9770e-04
Epoch 552/1500
Epoch 00551: val_loss improved from 0.00058 to 0.00057, saving model to my_model.h5
3s - loss: 7.3795e-04 - val_loss: 5.7487e-04
Epoch 553/1500
Epoch 00552: val_loss did not improve
3s - loss: 7.3197e-04 - val_loss: 6.1411e-04
Epoch 554/1500
Epoch 00553: val_loss did not improve
3s - loss: 7.2008e-04 - val_loss: 6.2161e-04
Epoch 555/1500
Epoch 00554: val_loss did not improve
3s - loss: 7.3523e-04 - val_loss: 5.9007e-04
Epoch 556/1500
Epoch 00555: val_loss did not improve
3s - loss: 7.7436e-04 - val_loss: 5.9786e-04
Epoch 557/1500
Epoch 00556: val_loss did not improve
3s - loss: 7.4300e-04 - val_loss: 5.9165e-04
Epoch 558/1500
Epoch 00557: val_loss did not improve
3s - loss: 7.2922e-04 - val_loss: 5.9156e-04
Epoch 559/1500
Epoch 00558: val_loss did not improve
3s - loss: 7.5536e-04 - val_loss: 5.8981e-04
Epoch 560/1500
Epoch 00559: val_loss did not improve
3s - loss: 7.2315e-04 - val_loss: 5.9359e-04
Epoch 561/1500
Epoch 00560: val_loss did not improve
3s - loss: 7.2098e-04 - val_loss: 5.8887e-04
Epoch 562/1500
Epoch 00561: val_loss did not improve
3s - loss: 7.4199e-04 - val_loss: 6.0212e-04
Epoch 563/1500
Epoch 00562: val_loss did not improve
3s - loss: 7.0801e-04 - val_loss: 5.9051e-04
Epoch 564/1500
Epoch 00563: val_loss did not improve
3s - loss: 7.4037e-04 - val_loss: 5.9454e-04
Epoch 565/1500
Epoch 00564: val_loss did not improve
3s - loss: 7.3215e-04 - val_loss: 6.0423e-04
Epoch 566/1500
Epoch 00565: val_loss did not improve
3s - loss: 7.1430e-04 - val_loss: 5.9747e-04
Epoch 567/1500
Epoch 00566: val_loss did not improve
3s - loss: 7.3282e-04 - val_loss: 5.8916e-04
Epoch 568/1500
Epoch 00567: val_loss did not improve
3s - loss: 7.3932e-04 - val_loss: 5.7881e-04
Epoch 569/1500
Epoch 00568: val_loss did not improve
3s - loss: 7.2093e-04 - val_loss: 6.0242e-04
Epoch 570/1500
Epoch 00569: val_loss did not improve
3s - loss: 7.1735e-04 - val_loss: 6.0007e-04
Epoch 571/1500
Epoch 00570: val_loss did not improve
3s - loss: 7.1818e-04 - val_loss: 5.7532e-04
Epoch 572/1500
Epoch 00571: val_loss did not improve
3s - loss: 7.1395e-04 - val_loss: 5.9425e-04
Epoch 573/1500
Epoch 00572: val_loss did not improve
3s - loss: 7.3164e-04 - val_loss: 5.9868e-04
Epoch 574/1500
Epoch 00573: val_loss did not improve
3s - loss: 7.3395e-04 - val_loss: 5.8265e-04
Epoch 575/1500
Epoch 00574: val_loss did not improve
3s - loss: 7.0692e-04 - val_loss: 5.8564e-04
Epoch 576/1500
Epoch 00575: val_loss did not improve
3s - loss: 7.1935e-04 - val_loss: 6.1330e-04
Epoch 577/1500
Epoch 00576: val_loss did not improve
3s - loss: 7.1902e-04 - val_loss: 5.9033e-04
Epoch 578/1500
Epoch 00577: val_loss did not improve
3s - loss: 6.9073e-04 - val_loss: 5.8240e-04
Epoch 579/1500
Epoch 00578: val_loss did not improve
3s - loss: 7.2442e-04 - val_loss: 5.8126e-04
Epoch 580/1500
Epoch 00579: val_loss did not improve
3s - loss: 7.0747e-04 - val_loss: 6.0657e-04
Epoch 581/1500
Epoch 00580: val_loss did not improve
3s - loss: 7.1383e-04 - val_loss: 6.1029e-04
Epoch 582/1500
Epoch 00581: val_loss did not improve
3s - loss: 7.3429e-04 - val_loss: 5.9232e-04
Epoch 583/1500
Epoch 00582: val_loss did not improve
3s - loss: 7.3793e-04 - val_loss: 5.7982e-04
Epoch 584/1500
Epoch 00583: val_loss did not improve
3s - loss: 7.1849e-04 - val_loss: 5.8701e-04
Epoch 585/1500
Epoch 00584: val_loss did not improve
3s - loss: 7.0521e-04 - val_loss: 5.8083e-04
Epoch 586/1500
Epoch 00585: val_loss did not improve
3s - loss: 6.9760e-04 - val_loss: 5.9761e-04
Epoch 587/1500
Epoch 00586: val_loss did not improve
3s - loss: 7.3430e-04 - val_loss: 5.9022e-04
Epoch 588/1500
Epoch 00587: val_loss did not improve
3s - loss: 7.1587e-04 - val_loss: 5.9375e-04
Epoch 589/1500
Epoch 00588: val_loss did not improve
3s - loss: 7.0030e-04 - val_loss: 5.9201e-04
Epoch 590/1500
Epoch 00589: val_loss did not improve
3s - loss: 7.2300e-04 - val_loss: 5.8688e-04
Epoch 591/1500
Epoch 00590: val_loss did not improve
3s - loss: 7.0172e-04 - val_loss: 5.8350e-04
Epoch 592/1500
Epoch 00591: val_loss did not improve
3s - loss: 6.9765e-04 - val_loss: 5.8037e-04
Epoch 593/1500
Epoch 00592: val_loss did not improve
3s - loss: 7.2383e-04 - val_loss: 6.1436e-04
Epoch 594/1500
Epoch 00593: val_loss did not improve
3s - loss: 7.6365e-04 - val_loss: 6.0280e-04
Epoch 595/1500
Epoch 00594: val_loss did not improve
3s - loss: 7.3896e-04 - val_loss: 5.9094e-04
Epoch 596/1500
Epoch 00595: val_loss improved from 0.00057 to 0.00057, saving model to my_model.h5
3s - loss: 7.2062e-04 - val_loss: 5.7388e-04
Epoch 597/1500
Epoch 00596: val_loss did not improve
3s - loss: 7.0685e-04 - val_loss: 5.9673e-04
Epoch 598/1500
Epoch 00597: val_loss did not improve
3s - loss: 7.1150e-04 - val_loss: 5.9554e-04
Epoch 599/1500
Epoch 00598: val_loss did not improve
3s - loss: 6.8963e-04 - val_loss: 5.9749e-04
Epoch 600/1500
Epoch 00599: val_loss improved from 0.00057 to 0.00057, saving model to my_model.h5
3s - loss: 7.1894e-04 - val_loss: 5.7198e-04
Epoch 601/1500
Epoch 00600: val_loss did not improve
3s - loss: 6.8531e-04 - val_loss: 5.7922e-04
Epoch 602/1500
Epoch 00601: val_loss did not improve
3s - loss: 7.2011e-04 - val_loss: 6.0985e-04
Epoch 603/1500
Epoch 00602: val_loss did not improve
3s - loss: 6.9244e-04 - val_loss: 5.7954e-04
Epoch 604/1500
Epoch 00603: val_loss did not improve
3s - loss: 6.9586e-04 - val_loss: 5.9481e-04
Epoch 605/1500
Epoch 00604: val_loss did not improve
3s - loss: 7.2295e-04 - val_loss: 5.7949e-04
Epoch 606/1500
Epoch 00605: val_loss did not improve
3s - loss: 6.9916e-04 - val_loss: 5.7423e-04
Epoch 607/1500
Epoch 00606: val_loss did not improve
3s - loss: 7.1341e-04 - val_loss: 5.8955e-04
Epoch 608/1500
Epoch 00607: val_loss did not improve
3s - loss: 6.9838e-04 - val_loss: 5.9946e-04
Epoch 609/1500
Epoch 00608: val_loss did not improve
3s - loss: 6.9772e-04 - val_loss: 5.9066e-04
Epoch 610/1500
Epoch 00609: val_loss did not improve
3s - loss: 7.1879e-04 - val_loss: 5.9204e-04
Epoch 611/1500
Epoch 00610: val_loss did not improve
3s - loss: 6.7775e-04 - val_loss: 5.8890e-04
Epoch 612/1500
Epoch 00611: val_loss did not improve
3s - loss: 7.0717e-04 - val_loss: 6.0596e-04
Epoch 613/1500
Epoch 00612: val_loss did not improve
3s - loss: 7.1044e-04 - val_loss: 6.1636e-04
Epoch 614/1500
Epoch 00613: val_loss did not improve
3s - loss: 7.0170e-04 - val_loss: 5.8320e-04
Epoch 615/1500
Epoch 00614: val_loss did not improve
3s - loss: 6.9146e-04 - val_loss: 5.7904e-04
Epoch 616/1500
Epoch 00615: val_loss did not improve
3s - loss: 6.6912e-04 - val_loss: 5.8014e-04
Epoch 617/1500
Epoch 00616: val_loss did not improve
3s - loss: 6.9221e-04 - val_loss: 5.8880e-04
Epoch 618/1500
Epoch 00617: val_loss did not improve
3s - loss: 7.2322e-04 - val_loss: 5.8112e-04
Epoch 619/1500
Epoch 00618: val_loss did not improve
3s - loss: 6.9011e-04 - val_loss: 5.8145e-04
Epoch 620/1500
Epoch 00619: val_loss improved from 0.00057 to 0.00056, saving model to my_model.h5
3s - loss: 7.2977e-04 - val_loss: 5.6415e-04
Epoch 621/1500
Epoch 00620: val_loss did not improve
3s - loss: 6.8781e-04 - val_loss: 5.9752e-04
Epoch 622/1500
Epoch 00621: val_loss did not improve
3s - loss: 6.8367e-04 - val_loss: 5.9345e-04
Epoch 623/1500
Epoch 00622: val_loss did not improve
13s - loss: 6.6454e-04 - val_loss: 5.8062e-04
Epoch 624/1500
Epoch 00623: val_loss did not improve
3s - loss: 6.9584e-04 - val_loss: 6.1158e-04
Epoch 625/1500
Epoch 00624: val_loss did not improve
3s - loss: 7.0728e-04 - val_loss: 5.7765e-04
Epoch 626/1500
Epoch 00625: val_loss did not improve
3s - loss: 7.0374e-04 - val_loss: 5.8024e-04
Epoch 627/1500
Epoch 00626: val_loss did not improve
3s - loss: 6.9085e-04 - val_loss: 5.6700e-04
Epoch 628/1500
Epoch 00627: val_loss did not improve
3s - loss: 7.2333e-04 - val_loss: 5.8641e-04
Epoch 629/1500
Epoch 00628: val_loss did not improve
3s - loss: 7.0071e-04 - val_loss: 5.7594e-04
Epoch 630/1500
Epoch 00629: val_loss did not improve
3s - loss: 6.7141e-04 - val_loss: 5.8368e-04
Epoch 631/1500
Epoch 00630: val_loss did not improve
3s - loss: 6.6832e-04 - val_loss: 5.7606e-04
Epoch 632/1500
Epoch 00631: val_loss did not improve
3s - loss: 6.7412e-04 - val_loss: 5.9929e-04
Epoch 633/1500
Epoch 00632: val_loss did not improve
3s - loss: 7.0784e-04 - val_loss: 5.8019e-04
Epoch 634/1500
Epoch 00633: val_loss did not improve
3s - loss: 6.9227e-04 - val_loss: 5.7733e-04
Epoch 635/1500
Epoch 00634: val_loss did not improve
3s - loss: 7.0408e-04 - val_loss: 5.8101e-04
Epoch 636/1500
Epoch 00635: val_loss did not improve
3s - loss: 6.6948e-04 - val_loss: 5.8363e-04
Epoch 637/1500
Epoch 00636: val_loss did not improve
3s - loss: 6.9456e-04 - val_loss: 5.7454e-04
Epoch 638/1500
Epoch 00637: val_loss did not improve
3s - loss: 6.7632e-04 - val_loss: 6.0566e-04
Epoch 639/1500
Epoch 00638: val_loss did not improve
3s - loss: 7.0045e-04 - val_loss: 5.7106e-04
Epoch 640/1500
Epoch 00639: val_loss did not improve
3s - loss: 6.8228e-04 - val_loss: 5.7291e-04
Epoch 641/1500
Epoch 00640: val_loss did not improve
3s - loss: 6.8175e-04 - val_loss: 5.6988e-04
Epoch 642/1500
Epoch 00641: val_loss did not improve
3s - loss: 6.7837e-04 - val_loss: 5.9551e-04
Epoch 643/1500
Epoch 00642: val_loss did not improve
3s - loss: 6.7969e-04 - val_loss: 5.8958e-04
Epoch 644/1500
Epoch 00643: val_loss improved from 0.00056 to 0.00056, saving model to my_model.h5
3s - loss: 6.7642e-04 - val_loss: 5.6074e-04
Epoch 645/1500
Epoch 00644: val_loss did not improve
3s - loss: 6.8551e-04 - val_loss: 5.6369e-04
Epoch 646/1500
Epoch 00645: val_loss did not improve
3s - loss: 6.9032e-04 - val_loss: 5.6947e-04
Epoch 647/1500
Epoch 00646: val_loss did not improve
3s - loss: 6.8604e-04 - val_loss: 5.6904e-04
Epoch 648/1500
Epoch 00647: val_loss did not improve
3s - loss: 6.9270e-04 - val_loss: 5.7612e-04
Epoch 649/1500
Epoch 00648: val_loss did not improve
3s - loss: 6.9520e-04 - val_loss: 5.8418e-04
Epoch 650/1500
Epoch 00649: val_loss did not improve
3s - loss: 6.8124e-04 - val_loss: 5.7952e-04
Epoch 651/1500
Epoch 00650: val_loss did not improve
3s - loss: 7.0284e-04 - val_loss: 5.7651e-04
Epoch 652/1500
Epoch 00651: val_loss did not improve
3s - loss: 6.9577e-04 - val_loss: 5.6931e-04
Epoch 653/1500
Epoch 00652: val_loss did not improve
3s - loss: 6.5377e-04 - val_loss: 6.0344e-04
Epoch 654/1500
Epoch 00653: val_loss did not improve
3s - loss: 6.8917e-04 - val_loss: 5.8346e-04
Epoch 655/1500
Epoch 00654: val_loss did not improve
3s - loss: 6.7145e-04 - val_loss: 5.9683e-04
Epoch 656/1500
Epoch 00655: val_loss did not improve
5s - loss: 6.9879e-04 - val_loss: 5.8395e-04
Epoch 657/1500
Epoch 00656: val_loss did not improve
4s - loss: 6.9751e-04 - val_loss: 5.7398e-04
Epoch 658/1500
Epoch 00657: val_loss did not improve
3s - loss: 6.9861e-04 - val_loss: 5.9901e-04
Epoch 659/1500
Epoch 00658: val_loss did not improve
3s - loss: 6.6509e-04 - val_loss: 5.8165e-04
Epoch 660/1500
Epoch 00659: val_loss did not improve
3s - loss: 6.8897e-04 - val_loss: 5.7024e-04
Epoch 661/1500
Epoch 00660: val_loss did not improve
3s - loss: 7.1320e-04 - val_loss: 5.7739e-04
Epoch 662/1500
Epoch 00661: val_loss did not improve
3s - loss: 6.8400e-04 - val_loss: 5.9286e-04
Epoch 663/1500
Epoch 00662: val_loss did not improve
3s - loss: 6.7843e-04 - val_loss: 5.6110e-04
Epoch 664/1500
Epoch 00663: val_loss did not improve
3s - loss: 6.6381e-04 - val_loss: 5.8240e-04
Epoch 665/1500
Epoch 00664: val_loss did not improve
3s - loss: 6.8608e-04 - val_loss: 5.8340e-04
Epoch 666/1500
Epoch 00665: val_loss did not improve
3s - loss: 6.8987e-04 - val_loss: 5.7435e-04
Epoch 667/1500
Epoch 00666: val_loss did not improve
3s - loss: 6.8894e-04 - val_loss: 5.7532e-04
Epoch 668/1500
Epoch 00667: val_loss did not improve
3s - loss: 6.9119e-04 - val_loss: 5.8271e-04
Epoch 669/1500
Epoch 00668: val_loss did not improve
3s - loss: 6.8796e-04 - val_loss: 5.7646e-04
Epoch 670/1500
Epoch 00669: val_loss did not improve
3s - loss: 6.6737e-04 - val_loss: 5.7460e-04
Epoch 671/1500
Epoch 00670: val_loss did not improve
3s - loss: 6.7654e-04 - val_loss: 5.6839e-04
Epoch 672/1500
Epoch 00671: val_loss did not improve
3s - loss: 6.9190e-04 - val_loss: 5.8140e-04
Epoch 673/1500
Epoch 00672: val_loss did not improve
3s - loss: 6.5724e-04 - val_loss: 6.0011e-04
Epoch 674/1500
Epoch 00673: val_loss did not improve
3s - loss: 6.7214e-04 - val_loss: 5.6394e-04
Epoch 675/1500
Epoch 00674: val_loss did not improve
3s - loss: 6.6754e-04 - val_loss: 5.8116e-04
Epoch 676/1500
Epoch 00675: val_loss did not improve
3s - loss: 6.6920e-04 - val_loss: 5.8296e-04
Epoch 677/1500
Epoch 00676: val_loss did not improve
3s - loss: 6.8480e-04 - val_loss: 5.6662e-04
Epoch 678/1500
Epoch 00677: val_loss improved from 0.00056 to 0.00056, saving model to my_model.h5
3s - loss: 6.8054e-04 - val_loss: 5.6011e-04
Epoch 679/1500
Epoch 00678: val_loss did not improve
3s - loss: 6.8318e-04 - val_loss: 5.6493e-04
Epoch 680/1500
Epoch 00679: val_loss did not improve
3s - loss: 6.9074e-04 - val_loss: 5.7111e-04
Epoch 681/1500
Epoch 00680: val_loss did not improve
3s - loss: 6.7876e-04 - val_loss: 5.9429e-04
Epoch 682/1500
Epoch 00681: val_loss did not improve
3s - loss: 6.9928e-04 - val_loss: 5.8013e-04
Epoch 683/1500
Epoch 00682: val_loss did not improve
3s - loss: 6.7819e-04 - val_loss: 5.7878e-04
Epoch 684/1500
Epoch 00683: val_loss did not improve
3s - loss: 6.9941e-04 - val_loss: 5.6865e-04
Epoch 685/1500
Epoch 00684: val_loss did not improve
3s - loss: 6.5696e-04 - val_loss: 5.6807e-04
Epoch 686/1500
Epoch 00685: val_loss did not improve
3s - loss: 6.8230e-04 - val_loss: 5.8138e-04
Epoch 687/1500
Epoch 00686: val_loss did not improve
3s - loss: 6.6794e-04 - val_loss: 5.8131e-04
Epoch 688/1500
Epoch 00687: val_loss did not improve
3s - loss: 6.5807e-04 - val_loss: 5.7315e-04
Epoch 689/1500
Epoch 00688: val_loss did not improve
3s - loss: 6.5023e-04 - val_loss: 5.9065e-04
Epoch 690/1500
Epoch 00689: val_loss did not improve
3s - loss: 6.7595e-04 - val_loss: 5.8236e-04
Epoch 691/1500
Epoch 00690: val_loss did not improve
3s - loss: 6.6655e-04 - val_loss: 5.8050e-04
Epoch 692/1500
Epoch 00691: val_loss did not improve
3s - loss: 6.7100e-04 - val_loss: 5.7629e-04
Epoch 693/1500
Epoch 00692: val_loss did not improve
3s - loss: 6.8671e-04 - val_loss: 5.6882e-04
Epoch 694/1500
Epoch 00693: val_loss did not improve
3s - loss: 6.7505e-04 - val_loss: 5.7098e-04
Epoch 695/1500
Epoch 00694: val_loss did not improve
3s - loss: 6.7294e-04 - val_loss: 5.8195e-04
Epoch 696/1500
Epoch 00695: val_loss did not improve
3s - loss: 6.5727e-04 - val_loss: 5.6680e-04
Epoch 697/1500
Epoch 00696: val_loss did not improve
3s - loss: 6.9172e-04 - val_loss: 5.6709e-04
Epoch 698/1500
Epoch 00697: val_loss improved from 0.00056 to 0.00055, saving model to my_model.h5
3s - loss: 6.6449e-04 - val_loss: 5.4941e-04
Epoch 699/1500
Epoch 00698: val_loss did not improve
3s - loss: 6.8425e-04 - val_loss: 5.8278e-04
Epoch 700/1500
Epoch 00699: val_loss did not improve
3s - loss: 6.4966e-04 - val_loss: 5.7313e-04
Epoch 701/1500
Epoch 00700: val_loss did not improve
3s - loss: 6.7997e-04 - val_loss: 5.7848e-04
Epoch 702/1500
Epoch 00701: val_loss did not improve
3s - loss: 6.6698e-04 - val_loss: 5.5989e-04
Epoch 703/1500
Epoch 00702: val_loss improved from 0.00055 to 0.00055, saving model to my_model.h5
3s - loss: 6.5920e-04 - val_loss: 5.4785e-04
Epoch 704/1500
Epoch 00703: val_loss did not improve
3s - loss: 6.8219e-04 - val_loss: 5.7017e-04
Epoch 705/1500
Epoch 00704: val_loss did not improve
3s - loss: 6.6719e-04 - val_loss: 5.8442e-04
Epoch 706/1500
Epoch 00705: val_loss did not improve
3s - loss: 6.7308e-04 - val_loss: 6.1130e-04
Epoch 707/1500
Epoch 00706: val_loss did not improve
3s - loss: 6.6420e-04 - val_loss: 5.7542e-04
Epoch 708/1500
Epoch 00707: val_loss did not improve
3s - loss: 6.8574e-04 - val_loss: 5.4920e-04
Epoch 709/1500
Epoch 00708: val_loss did not improve
3s - loss: 6.6993e-04 - val_loss: 5.6686e-04
Epoch 710/1500
Epoch 00709: val_loss did not improve
3s - loss: 6.5595e-04 - val_loss: 5.7982e-04
Epoch 711/1500
Epoch 00710: val_loss did not improve
3s - loss: 6.6518e-04 - val_loss: 5.7183e-04
Epoch 712/1500
Epoch 00711: val_loss did not improve
3s - loss: 6.7309e-04 - val_loss: 5.9168e-04
Epoch 713/1500
Epoch 00712: val_loss did not improve
3s - loss: 6.8212e-04 - val_loss: 5.8174e-04
Epoch 714/1500
Epoch 00713: val_loss did not improve
3s - loss: 6.5571e-04 - val_loss: 5.7450e-04
Epoch 715/1500
Epoch 00714: val_loss did not improve
3s - loss: 6.4792e-04 - val_loss: 5.6866e-04
Epoch 716/1500
Epoch 00715: val_loss did not improve
3s - loss: 6.2750e-04 - val_loss: 5.6985e-04
Epoch 717/1500
Epoch 00716: val_loss did not improve
3s - loss: 6.8531e-04 - val_loss: 5.5254e-04
Epoch 718/1500
Epoch 00717: val_loss did not improve
3s - loss: 6.7377e-04 - val_loss: 5.8450e-04
Epoch 719/1500
Epoch 00718: val_loss did not improve
3s - loss: 6.5967e-04 - val_loss: 6.0935e-04
Epoch 720/1500
Epoch 00719: val_loss did not improve
3s - loss: 6.3998e-04 - val_loss: 5.8278e-04
Epoch 721/1500
Epoch 00720: val_loss did not improve
3s - loss: 6.5522e-04 - val_loss: 5.7495e-04
Epoch 722/1500
Epoch 00721: val_loss did not improve
3s - loss: 6.6066e-04 - val_loss: 5.8719e-04
Epoch 723/1500
Epoch 00722: val_loss did not improve
3s - loss: 6.5085e-04 - val_loss: 5.6324e-04
Epoch 724/1500
Epoch 00723: val_loss did not improve
3s - loss: 6.4910e-04 - val_loss: 5.6526e-04
Epoch 725/1500
Epoch 00724: val_loss did not improve
3s - loss: 6.4829e-04 - val_loss: 5.9165e-04
Epoch 726/1500
Epoch 00725: val_loss did not improve
3s - loss: 6.7656e-04 - val_loss: 5.7286e-04
Epoch 727/1500
Epoch 00726: val_loss did not improve
3s - loss: 6.4850e-04 - val_loss: 5.6615e-04
Epoch 728/1500
Epoch 00727: val_loss did not improve
3s - loss: 6.4050e-04 - val_loss: 5.7059e-04
Epoch 729/1500
Epoch 00728: val_loss did not improve
3s - loss: 6.5012e-04 - val_loss: 5.7775e-04
Epoch 730/1500
Epoch 00729: val_loss did not improve
3s - loss: 6.8571e-04 - val_loss: 5.8305e-04
Epoch 731/1500
Epoch 00730: val_loss did not improve
3s - loss: 6.4821e-04 - val_loss: 5.6859e-04
Epoch 732/1500
Epoch 00731: val_loss did not improve
3s - loss: 6.5561e-04 - val_loss: 5.6447e-04
Epoch 733/1500
Epoch 00732: val_loss did not improve
3s - loss: 6.6676e-04 - val_loss: 5.9064e-04
Epoch 734/1500
Epoch 00733: val_loss did not improve
3s - loss: 6.5064e-04 - val_loss: 5.8100e-04
Epoch 735/1500
Epoch 00734: val_loss did not improve
3s - loss: 6.2843e-04 - val_loss: 5.7289e-04
Epoch 736/1500
Epoch 00735: val_loss did not improve
3s - loss: 6.3995e-04 - val_loss: 5.5421e-04
Epoch 737/1500
Epoch 00736: val_loss did not improve
3s - loss: 6.4394e-04 - val_loss: 5.7092e-04
Epoch 738/1500
Epoch 00737: val_loss did not improve
3s - loss: 6.6071e-04 - val_loss: 5.5930e-04
Epoch 739/1500
Epoch 00738: val_loss did not improve
3s - loss: 6.6580e-04 - val_loss: 5.5558e-04
Epoch 740/1500
Epoch 00739: val_loss did not improve
3s - loss: 6.5479e-04 - val_loss: 5.6991e-04
Epoch 741/1500
Epoch 00740: val_loss did not improve
3s - loss: 6.6137e-04 - val_loss: 6.3084e-04
Epoch 742/1500
Epoch 00741: val_loss did not improve
3s - loss: 6.6732e-04 - val_loss: 5.7429e-04
Epoch 743/1500
Epoch 00742: val_loss improved from 0.00055 to 0.00054, saving model to my_model.h5
3s - loss: 6.3397e-04 - val_loss: 5.4377e-04
Epoch 744/1500
Epoch 00743: val_loss did not improve
3s - loss: 6.7527e-04 - val_loss: 5.5223e-04
Epoch 745/1500
Epoch 00744: val_loss did not improve
3s - loss: 6.4887e-04 - val_loss: 5.6598e-04
Epoch 746/1500
Epoch 00745: val_loss did not improve
3s - loss: 6.2877e-04 - val_loss: 5.5864e-04
Epoch 747/1500
Epoch 00746: val_loss did not improve
3s - loss: 6.5809e-04 - val_loss: 5.8383e-04
Epoch 748/1500
Epoch 00747: val_loss did not improve
3s - loss: 6.3933e-04 - val_loss: 5.6446e-04
Epoch 749/1500
Epoch 00748: val_loss did not improve
3s - loss: 6.4083e-04 - val_loss: 5.5601e-04
Epoch 750/1500
Epoch 00749: val_loss did not improve
3s - loss: 6.5810e-04 - val_loss: 5.6303e-04
Epoch 751/1500
Epoch 00750: val_loss did not improve
3s - loss: 6.4630e-04 - val_loss: 5.5242e-04
Epoch 752/1500
Epoch 00751: val_loss did not improve
3s - loss: 6.3994e-04 - val_loss: 5.7002e-04
Epoch 753/1500
Epoch 00752: val_loss did not improve
3s - loss: 6.6037e-04 - val_loss: 5.7713e-04
Epoch 754/1500
Epoch 00753: val_loss did not improve
3s - loss: 6.6853e-04 - val_loss: 5.5976e-04
Epoch 755/1500
Epoch 00754: val_loss did not improve
3s - loss: 6.6153e-04 - val_loss: 5.6486e-04
Epoch 756/1500
Epoch 00755: val_loss did not improve
3s - loss: 6.4723e-04 - val_loss: 5.5537e-04
Epoch 757/1500
Epoch 00756: val_loss did not improve
3s - loss: 6.6341e-04 - val_loss: 5.5756e-04
Epoch 758/1500
Epoch 00757: val_loss did not improve
3s - loss: 6.5155e-04 - val_loss: 5.9181e-04
Epoch 759/1500
Epoch 00758: val_loss did not improve
3s - loss: 6.5715e-04 - val_loss: 5.6749e-04
Epoch 760/1500
Epoch 00759: val_loss did not improve
3s - loss: 6.5071e-04 - val_loss: 5.5322e-04
Epoch 761/1500
Epoch 00760: val_loss did not improve
3s - loss: 6.5784e-04 - val_loss: 5.6342e-04
Epoch 762/1500
Epoch 00761: val_loss did not improve
3s - loss: 6.5796e-04 - val_loss: 5.6701e-04
Epoch 763/1500
Epoch 00762: val_loss did not improve
3s - loss: 6.4309e-04 - val_loss: 5.7443e-04
Epoch 764/1500
Epoch 00763: val_loss did not improve
3s - loss: 6.6184e-04 - val_loss: 5.6613e-04
Epoch 765/1500
Epoch 00764: val_loss did not improve
3s - loss: 6.8179e-04 - val_loss: 5.7331e-04
Epoch 766/1500
Epoch 00765: val_loss did not improve
3s - loss: 6.4425e-04 - val_loss: 5.5241e-04
Epoch 767/1500
Epoch 00766: val_loss did not improve
3s - loss: 6.4770e-04 - val_loss: 5.6191e-04
Epoch 768/1500
Epoch 00767: val_loss did not improve
3s - loss: 6.5538e-04 - val_loss: 5.7486e-04
Epoch 769/1500
Epoch 00768: val_loss did not improve
3s - loss: 6.3967e-04 - val_loss: 5.5635e-04
Epoch 770/1500
Epoch 00769: val_loss did not improve
3s - loss: 6.4767e-04 - val_loss: 5.6618e-04
Epoch 771/1500
Epoch 00770: val_loss did not improve
3s - loss: 6.6052e-04 - val_loss: 5.6988e-04
Epoch 772/1500
Epoch 00771: val_loss did not improve
3s - loss: 6.3220e-04 - val_loss: 5.6310e-04
Epoch 773/1500
Epoch 00772: val_loss did not improve
3s - loss: 6.4333e-04 - val_loss: 5.8622e-04
Epoch 774/1500
Epoch 00773: val_loss did not improve
3s - loss: 6.6038e-04 - val_loss: 5.5526e-04
Epoch 775/1500
Epoch 00774: val_loss did not improve
3s - loss: 6.5805e-04 - val_loss: 5.6513e-04
Epoch 776/1500
Epoch 00775: val_loss did not improve
3s - loss: 6.4741e-04 - val_loss: 5.5026e-04
Epoch 777/1500
Epoch 00776: val_loss improved from 0.00054 to 0.00054, saving model to my_model.h5
3s - loss: 6.5836e-04 - val_loss: 5.4362e-04
Epoch 778/1500
Epoch 00777: val_loss did not improve
3s - loss: 6.4898e-04 - val_loss: 5.6013e-04
Epoch 779/1500
Epoch 00778: val_loss did not improve
3s - loss: 6.2119e-04 - val_loss: 5.5471e-04
Epoch 780/1500
Epoch 00779: val_loss did not improve
3s - loss: 6.4517e-04 - val_loss: 5.6493e-04
Epoch 781/1500
Epoch 00780: val_loss did not improve
3s - loss: 6.5201e-04 - val_loss: 5.6641e-04
Epoch 782/1500
Epoch 00781: val_loss did not improve
3s - loss: 6.2003e-04 - val_loss: 5.6770e-04
Epoch 783/1500
Epoch 00782: val_loss did not improve
3s - loss: 6.4404e-04 - val_loss: 5.8866e-04
Epoch 784/1500
Epoch 00783: val_loss did not improve
3s - loss: 6.5036e-04 - val_loss: 5.6120e-04
Epoch 785/1500
Epoch 00784: val_loss did not improve
3s - loss: 6.6975e-04 - val_loss: 5.7674e-04
Epoch 786/1500
Epoch 00785: val_loss did not improve
3s - loss: 6.5846e-04 - val_loss: 5.8209e-04
Epoch 787/1500
Epoch 00786: val_loss did not improve
3s - loss: 6.6788e-04 - val_loss: 5.6323e-04
Epoch 788/1500
Epoch 00787: val_loss did not improve
3s - loss: 6.4215e-04 - val_loss: 5.6794e-04
Epoch 789/1500
Epoch 00788: val_loss did not improve
3s - loss: 6.5329e-04 - val_loss: 5.7971e-04
Epoch 790/1500
Epoch 00789: val_loss did not improve
3s - loss: 6.4658e-04 - val_loss: 5.5852e-04
Epoch 791/1500
Epoch 00790: val_loss did not improve
3s - loss: 6.5861e-04 - val_loss: 5.6471e-04
Epoch 792/1500
Epoch 00791: val_loss did not improve
3s - loss: 6.6385e-04 - val_loss: 5.8236e-04
Epoch 793/1500
Epoch 00792: val_loss did not improve
3s - loss: 6.2309e-04 - val_loss: 5.7197e-04
Epoch 794/1500
Epoch 00793: val_loss did not improve
3s - loss: 6.5199e-04 - val_loss: 5.7137e-04
Epoch 795/1500
Epoch 00794: val_loss did not improve
3s - loss: 6.6351e-04 - val_loss: 5.7829e-04
Epoch 796/1500
Epoch 00795: val_loss did not improve
3s - loss: 6.4682e-04 - val_loss: 5.5156e-04
Epoch 797/1500
Epoch 00796: val_loss did not improve
3s - loss: 6.5980e-04 - val_loss: 5.6998e-04
Epoch 798/1500
Epoch 00797: val_loss did not improve
3s - loss: 6.5797e-04 - val_loss: 5.6981e-04
Epoch 799/1500
Epoch 00798: val_loss did not improve
3s - loss: 6.5186e-04 - val_loss: 5.6129e-04
Epoch 800/1500
Epoch 00799: val_loss did not improve
3s - loss: 6.6019e-04 - val_loss: 5.6513e-04
Epoch 801/1500
Epoch 00800: val_loss did not improve
3s - loss: 6.8646e-04 - val_loss: 5.6855e-04
Epoch 802/1500
Epoch 00801: val_loss did not improve
3s - loss: 6.7806e-04 - val_loss: 5.8655e-04
Epoch 803/1500
Epoch 00802: val_loss did not improve
3s - loss: 6.4582e-04 - val_loss: 5.7638e-04
Epoch 804/1500
Epoch 00803: val_loss did not improve
3s - loss: 6.5630e-04 - val_loss: 5.6208e-04
Epoch 805/1500
Epoch 00804: val_loss did not improve
3s - loss: 6.4133e-04 - val_loss: 5.8852e-04
Epoch 806/1500
Epoch 00805: val_loss did not improve
3s - loss: 6.4771e-04 - val_loss: 5.4946e-04
Epoch 807/1500
Epoch 00806: val_loss did not improve
3s - loss: 6.4350e-04 - val_loss: 5.7798e-04
Epoch 808/1500
Epoch 00807: val_loss did not improve
3s - loss: 6.4672e-04 - val_loss: 6.0962e-04
Epoch 809/1500
Epoch 00808: val_loss did not improve
3s - loss: 6.6597e-04 - val_loss: 5.7197e-04
Epoch 810/1500
Epoch 00809: val_loss did not improve
3s - loss: 6.3617e-04 - val_loss: 5.5172e-04
Epoch 811/1500
Epoch 00810: val_loss did not improve
3s - loss: 6.3646e-04 - val_loss: 5.5208e-04
Epoch 812/1500
Epoch 00811: val_loss did not improve
3s - loss: 6.3155e-04 - val_loss: 5.8652e-04
Epoch 813/1500
Epoch 00812: val_loss did not improve
3s - loss: 6.3540e-04 - val_loss: 5.6877e-04
Epoch 814/1500
Epoch 00813: val_loss did not improve
3s - loss: 6.2472e-04 - val_loss: 5.7290e-04
Epoch 815/1500
Epoch 00814: val_loss did not improve
3s - loss: 6.4519e-04 - val_loss: 5.7586e-04
Epoch 816/1500
Epoch 00815: val_loss did not improve
3s - loss: 6.3181e-04 - val_loss: 5.6750e-04
Epoch 817/1500
Epoch 00816: val_loss did not improve
3s - loss: 6.3803e-04 - val_loss: 5.6666e-04
Epoch 818/1500
Epoch 00817: val_loss did not improve
3s - loss: 6.4757e-04 - val_loss: 5.7978e-04
Epoch 819/1500
Epoch 00818: val_loss did not improve
3s - loss: 6.4149e-04 - val_loss: 5.5715e-04
Epoch 820/1500
Epoch 00819: val_loss did not improve
3s - loss: 6.3853e-04 - val_loss: 5.7752e-04
Epoch 821/1500
Epoch 00820: val_loss did not improve
3s - loss: 6.1356e-04 - val_loss: 5.7650e-04
Epoch 822/1500
Epoch 00821: val_loss did not improve
3s - loss: 6.3194e-04 - val_loss: 5.6973e-04
Epoch 823/1500
Epoch 00822: val_loss did not improve
3s - loss: 6.4663e-04 - val_loss: 5.7531e-04
Epoch 824/1500
Epoch 00823: val_loss did not improve
3s - loss: 6.3456e-04 - val_loss: 5.8781e-04
Epoch 825/1500
Epoch 00824: val_loss did not improve
3s - loss: 6.3577e-04 - val_loss: 5.6253e-04
Epoch 826/1500
Epoch 00825: val_loss did not improve
3s - loss: 6.4416e-04 - val_loss: 5.5906e-04
Epoch 827/1500
Epoch 00826: val_loss did not improve
3s - loss: 6.4321e-04 - val_loss: 5.5996e-04
Epoch 828/1500
Epoch 00827: val_loss did not improve
3s - loss: 6.0598e-04 - val_loss: 6.0425e-04
Epoch 829/1500
Epoch 00828: val_loss did not improve
3s - loss: 6.4377e-04 - val_loss: 5.8773e-04
Epoch 830/1500
Epoch 00829: val_loss did not improve
3s - loss: 6.6572e-04 - val_loss: 5.6471e-04
Epoch 831/1500
Epoch 00830: val_loss did not improve
3s - loss: 6.4128e-04 - val_loss: 5.5559e-04
Epoch 832/1500
Epoch 00831: val_loss did not improve
3s - loss: 6.5493e-04 - val_loss: 5.5232e-04
Epoch 833/1500
Epoch 00832: val_loss did not improve
3s - loss: 6.2976e-04 - val_loss: 5.5103e-04
Epoch 834/1500
Epoch 00833: val_loss did not improve
3s - loss: 6.2826e-04 - val_loss: 5.6567e-04
Epoch 835/1500
Epoch 00834: val_loss did not improve
3s - loss: 6.5865e-04 - val_loss: 5.6900e-04
Epoch 836/1500
Epoch 00835: val_loss did not improve
3s - loss: 6.3584e-04 - val_loss: 6.0793e-04
Epoch 837/1500
Epoch 00836: val_loss did not improve
3s - loss: 6.3595e-04 - val_loss: 5.7202e-04
Epoch 838/1500
Epoch 00837: val_loss did not improve
3s - loss: 6.4219e-04 - val_loss: 5.7627e-04
Epoch 839/1500
Epoch 00838: val_loss did not improve
3s - loss: 6.3337e-04 - val_loss: 5.6956e-04
Epoch 840/1500
Epoch 00839: val_loss did not improve
3s - loss: 6.3740e-04 - val_loss: 5.5998e-04
Epoch 841/1500
Epoch 00840: val_loss did not improve
3s - loss: 6.2665e-04 - val_loss: 5.5451e-04
Epoch 842/1500
Epoch 00841: val_loss did not improve
3s - loss: 6.1024e-04 - val_loss: 5.5460e-04
Epoch 843/1500
Epoch 00842: val_loss did not improve
3s - loss: 6.3332e-04 - val_loss: 5.5796e-04
Epoch 844/1500
Epoch 00843: val_loss did not improve
3s - loss: 6.3809e-04 - val_loss: 5.4949e-04
Epoch 845/1500
Epoch 00844: val_loss did not improve
3s - loss: 6.3107e-04 - val_loss: 5.6219e-04
Epoch 846/1500
Epoch 00845: val_loss did not improve
3s - loss: 6.2396e-04 - val_loss: 5.5463e-04
Epoch 847/1500
Epoch 00846: val_loss did not improve
3s - loss: 6.4518e-04 - val_loss: 5.5889e-04
Epoch 848/1500
Epoch 00847: val_loss did not improve
3s - loss: 6.3719e-04 - val_loss: 5.6338e-04
Epoch 849/1500
Epoch 00848: val_loss did not improve
3s - loss: 6.3053e-04 - val_loss: 5.8372e-04
Epoch 850/1500
Epoch 00849: val_loss did not improve
3s - loss: 6.3145e-04 - val_loss: 5.6760e-04
Epoch 851/1500
Epoch 00850: val_loss did not improve
3s - loss: 6.4923e-04 - val_loss: 5.4965e-04
Epoch 852/1500
Epoch 00851: val_loss did not improve
3s - loss: 6.3542e-04 - val_loss: 5.6843e-04
Epoch 853/1500
Epoch 00852: val_loss did not improve
3s - loss: 6.5624e-04 - val_loss: 6.0234e-04
Epoch 854/1500
Epoch 00853: val_loss did not improve
3s - loss: 6.3482e-04 - val_loss: 5.7615e-04
Epoch 855/1500
Epoch 00854: val_loss did not improve
3s - loss: 6.4765e-04 - val_loss: 5.7678e-04
Epoch 856/1500
Epoch 00855: val_loss did not improve
3s - loss: 6.6852e-04 - val_loss: 5.5667e-04
Epoch 857/1500
Epoch 00856: val_loss did not improve
3s - loss: 6.1783e-04 - val_loss: 5.6732e-04
Epoch 858/1500
Epoch 00857: val_loss did not improve
3s - loss: 6.4111e-04 - val_loss: 5.6694e-04
Epoch 859/1500
Epoch 00858: val_loss did not improve
3s - loss: 6.3686e-04 - val_loss: 5.6332e-04
Epoch 860/1500
Epoch 00859: val_loss improved from 0.00054 to 0.00053, saving model to my_model.h5
3s - loss: 6.4046e-04 - val_loss: 5.3471e-04
Epoch 861/1500
Epoch 00860: val_loss did not improve
3s - loss: 6.4294e-04 - val_loss: 5.8044e-04
Epoch 862/1500
Epoch 00861: val_loss did not improve
3s - loss: 6.3675e-04 - val_loss: 6.1895e-04
Epoch 863/1500
Epoch 00862: val_loss did not improve
3s - loss: 6.4662e-04 - val_loss: 5.6190e-04
Epoch 864/1500
Epoch 00863: val_loss did not improve
3s - loss: 6.3790e-04 - val_loss: 5.8408e-04
Epoch 865/1500
Epoch 00864: val_loss did not improve
3s - loss: 6.4556e-04 - val_loss: 5.7686e-04
Epoch 866/1500
Epoch 00865: val_loss did not improve
3s - loss: 6.4972e-04 - val_loss: 5.6489e-04
Epoch 867/1500
Epoch 00866: val_loss did not improve
3s - loss: 6.1841e-04 - val_loss: 5.4463e-04
Epoch 868/1500
Epoch 00867: val_loss did not improve
3s - loss: 6.2709e-04 - val_loss: 5.6159e-04
Epoch 869/1500
Epoch 00868: val_loss did not improve
3s - loss: 6.3108e-04 - val_loss: 5.7421e-04
Epoch 870/1500
Epoch 00869: val_loss did not improve
3s - loss: 6.3040e-04 - val_loss: 5.6612e-04
Epoch 871/1500
Epoch 00870: val_loss did not improve
3s - loss: 6.2230e-04 - val_loss: 5.4742e-04
Epoch 872/1500
Epoch 00871: val_loss did not improve
3s - loss: 6.2669e-04 - val_loss: 5.8128e-04
Epoch 873/1500
Epoch 00872: val_loss did not improve
3s - loss: 6.0935e-04 - val_loss: 5.6282e-04
Epoch 874/1500
Epoch 00873: val_loss did not improve
3s - loss: 6.2516e-04 - val_loss: 5.4649e-04
Epoch 875/1500
Epoch 00874: val_loss did not improve
3s - loss: 6.2972e-04 - val_loss: 5.5954e-04
Epoch 876/1500
Epoch 00875: val_loss did not improve
3s - loss: 6.3871e-04 - val_loss: 5.6630e-04
Epoch 877/1500
Epoch 00876: val_loss did not improve
3s - loss: 6.2319e-04 - val_loss: 5.5010e-04
Epoch 878/1500
Epoch 00877: val_loss did not improve
3s - loss: 6.0847e-04 - val_loss: 5.5271e-04
Epoch 879/1500
Epoch 00878: val_loss did not improve
3s - loss: 6.2371e-04 - val_loss: 5.7835e-04
Epoch 880/1500
Epoch 00879: val_loss did not improve
3s - loss: 6.3237e-04 - val_loss: 5.5292e-04
Epoch 881/1500
Epoch 00880: val_loss did not improve
3s - loss: 6.2458e-04 - val_loss: 5.6522e-04
Epoch 882/1500
Epoch 00881: val_loss did not improve
3s - loss: 6.1262e-04 - val_loss: 5.6129e-04
Epoch 883/1500
Epoch 00882: val_loss did not improve
3s - loss: 6.2598e-04 - val_loss: 5.4707e-04
Epoch 884/1500
Epoch 00883: val_loss did not improve
3s - loss: 6.1308e-04 - val_loss: 5.6005e-04
Epoch 885/1500
Epoch 00884: val_loss did not improve
3s - loss: 6.3829e-04 - val_loss: 5.6875e-04
Epoch 886/1500
Epoch 00885: val_loss did not improve
3s - loss: 6.0483e-04 - val_loss: 5.4172e-04
Epoch 887/1500
Epoch 00886: val_loss did not improve
3s - loss: 6.2951e-04 - val_loss: 5.8658e-04
Epoch 888/1500
Epoch 00887: val_loss did not improve
3s - loss: 6.4356e-04 - val_loss: 5.7459e-04
Epoch 889/1500
Epoch 00888: val_loss did not improve
3s - loss: 6.3724e-04 - val_loss: 5.6265e-04
Epoch 890/1500
Epoch 00889: val_loss did not improve
3s - loss: 6.3930e-04 - val_loss: 5.6192e-04
Epoch 891/1500
Epoch 00890: val_loss did not improve
3s - loss: 6.3441e-04 - val_loss: 5.7022e-04
Epoch 892/1500
Epoch 00891: val_loss did not improve
3s - loss: 6.2518e-04 - val_loss: 5.6439e-04
Epoch 893/1500
Epoch 00892: val_loss did not improve
3s - loss: 6.3217e-04 - val_loss: 5.6188e-04
Epoch 894/1500
Epoch 00893: val_loss did not improve
3s - loss: 6.1525e-04 - val_loss: 5.5032e-04
Epoch 895/1500
Epoch 00894: val_loss did not improve
3s - loss: 6.2963e-04 - val_loss: 5.5365e-04
Epoch 896/1500
Epoch 00895: val_loss did not improve
3s - loss: 6.3983e-04 - val_loss: 5.6736e-04
Epoch 897/1500
Epoch 00896: val_loss did not improve
3s - loss: 6.4054e-04 - val_loss: 6.0026e-04
Epoch 898/1500
Epoch 00897: val_loss did not improve
3s - loss: 6.4930e-04 - val_loss: 5.8584e-04
Epoch 899/1500
Epoch 00898: val_loss did not improve
3s - loss: 6.5346e-04 - val_loss: 5.6447e-04
Epoch 900/1500
Epoch 00899: val_loss did not improve
3s - loss: 6.2066e-04 - val_loss: 5.6840e-04
Epoch 901/1500
Epoch 00900: val_loss did not improve
3s - loss: 6.3331e-04 - val_loss: 5.6869e-04
Epoch 902/1500
Epoch 00901: val_loss did not improve
3s - loss: 6.5813e-04 - val_loss: 5.8137e-04
Epoch 903/1500
Epoch 00902: val_loss did not improve
3s - loss: 6.3776e-04 - val_loss: 5.5438e-04
Epoch 904/1500
Epoch 00903: val_loss did not improve
3s - loss: 6.2509e-04 - val_loss: 5.6232e-04
Epoch 905/1500
Epoch 00904: val_loss did not improve
3s - loss: 6.2784e-04 - val_loss: 5.8167e-04
Epoch 906/1500
Epoch 00905: val_loss did not improve
3s - loss: 6.3879e-04 - val_loss: 5.5957e-04
Epoch 907/1500
Epoch 00906: val_loss did not improve
3s - loss: 6.3588e-04 - val_loss: 5.5405e-04
Epoch 908/1500
Epoch 00907: val_loss did not improve
3s - loss: 6.2835e-04 - val_loss: 5.6339e-04
Epoch 909/1500
Epoch 00908: val_loss did not improve
3s - loss: 6.3256e-04 - val_loss: 5.9891e-04
Epoch 910/1500
Epoch 00909: val_loss did not improve
3s - loss: 6.2233e-04 - val_loss: 5.5429e-04
Epoch 911/1500
Epoch 00910: val_loss did not improve
3s - loss: 6.2953e-04 - val_loss: 5.5595e-04
Epoch 912/1500
Epoch 00911: val_loss did not improve
3s - loss: 6.1656e-04 - val_loss: 5.8730e-04
Epoch 913/1500
Epoch 00912: val_loss did not improve
3s - loss: 6.1655e-04 - val_loss: 5.7443e-04
Epoch 914/1500
Epoch 00913: val_loss did not improve
3s - loss: 6.1282e-04 - val_loss: 5.7150e-04
Epoch 915/1500
Epoch 00914: val_loss did not improve
3s - loss: 6.3127e-04 - val_loss: 5.7149e-04
Epoch 916/1500
Epoch 00915: val_loss did not improve
3s - loss: 6.2338e-04 - val_loss: 5.7235e-04
Epoch 917/1500
Epoch 00916: val_loss did not improve
3s - loss: 6.1486e-04 - val_loss: 5.6437e-04
Epoch 918/1500
Epoch 00917: val_loss did not improve
3s - loss: 6.1690e-04 - val_loss: 5.7411e-04
Epoch 919/1500
Epoch 00918: val_loss did not improve
3s - loss: 6.2545e-04 - val_loss: 5.5957e-04
Epoch 920/1500
Epoch 00919: val_loss did not improve
3s - loss: 5.9694e-04 - val_loss: 5.6925e-04
Epoch 921/1500
Epoch 00920: val_loss did not improve
3s - loss: 6.0541e-04 - val_loss: 5.5920e-04
Epoch 922/1500
Epoch 00921: val_loss did not improve
3s - loss: 6.3427e-04 - val_loss: 5.5344e-04
Epoch 923/1500
Epoch 00922: val_loss did not improve
3s - loss: 5.9729e-04 - val_loss: 5.6814e-04
Epoch 924/1500
Epoch 00923: val_loss did not improve
3s - loss: 6.0985e-04 - val_loss: 5.5978e-04
Epoch 925/1500
Epoch 00924: val_loss did not improve
3s - loss: 6.2961e-04 - val_loss: 5.6828e-04
Epoch 926/1500
Epoch 00925: val_loss did not improve
3s - loss: 6.0977e-04 - val_loss: 5.5020e-04
Epoch 927/1500
Epoch 00926: val_loss did not improve
3s - loss: 6.1682e-04 - val_loss: 5.4824e-04
Epoch 928/1500
Epoch 00927: val_loss did not improve
3s - loss: 6.2350e-04 - val_loss: 5.5053e-04
Epoch 929/1500
Epoch 00928: val_loss did not improve
3s - loss: 6.4187e-04 - val_loss: 5.8743e-04
Epoch 930/1500
Epoch 00929: val_loss did not improve
3s - loss: 6.4018e-04 - val_loss: 5.6747e-04
Epoch 931/1500
Epoch 00930: val_loss did not improve
8s - loss: 6.1052e-04 - val_loss: 5.6256e-04
Epoch 932/1500
Epoch 00931: val_loss did not improve
5s - loss: 6.3283e-04 - val_loss: 5.7243e-04
Epoch 933/1500
Epoch 00932: val_loss did not improve
3s - loss: 6.2976e-04 - val_loss: 5.5210e-04
Epoch 934/1500
Epoch 00933: val_loss did not improve
3s - loss: 6.1637e-04 - val_loss: 5.5668e-04
Epoch 935/1500
Epoch 00934: val_loss did not improve
3s - loss: 6.2933e-04 - val_loss: 5.4547e-04
Epoch 936/1500
Epoch 00935: val_loss did not improve
3s - loss: 6.1006e-04 - val_loss: 5.5818e-04
Epoch 937/1500
Epoch 00936: val_loss did not improve
3s - loss: 6.2978e-04 - val_loss: 5.6386e-04
Epoch 938/1500
Epoch 00937: val_loss did not improve
3s - loss: 6.2342e-04 - val_loss: 5.7326e-04
Epoch 939/1500
Epoch 00938: val_loss did not improve
3s - loss: 6.1250e-04 - val_loss: 5.5583e-04
Epoch 940/1500
Epoch 00939: val_loss did not improve
3s - loss: 6.3371e-04 - val_loss: 5.6911e-04
Epoch 941/1500
Epoch 00940: val_loss did not improve
3s - loss: 6.1291e-04 - val_loss: 5.7162e-04
Epoch 942/1500
Epoch 00941: val_loss did not improve
3s - loss: 6.1893e-04 - val_loss: 5.5642e-04
Epoch 943/1500
Epoch 00942: val_loss did not improve
3s - loss: 6.3269e-04 - val_loss: 5.6522e-04
Epoch 944/1500
Epoch 00943: val_loss did not improve
3s - loss: 6.2142e-04 - val_loss: 5.3858e-04
Epoch 945/1500
Epoch 00944: val_loss did not improve
3s - loss: 5.9011e-04 - val_loss: 5.6258e-04
Epoch 946/1500
Epoch 00945: val_loss did not improve
3s - loss: 6.4053e-04 - val_loss: 5.4647e-04
Epoch 947/1500
Epoch 00946: val_loss did not improve
3s - loss: 6.3403e-04 - val_loss: 5.4511e-04
Epoch 948/1500
Epoch 00947: val_loss did not improve
3s - loss: 5.8546e-04 - val_loss: 5.7404e-04
Epoch 949/1500
Epoch 00948: val_loss did not improve
3s - loss: 6.0601e-04 - val_loss: 5.5912e-04
Epoch 950/1500
Epoch 00949: val_loss did not improve
3s - loss: 5.9659e-04 - val_loss: 5.6700e-04
Epoch 951/1500
Epoch 00950: val_loss did not improve
3s - loss: 6.1316e-04 - val_loss: 5.5321e-04
Epoch 952/1500
Epoch 00951: val_loss did not improve
3s - loss: 6.2424e-04 - val_loss: 5.8079e-04
Epoch 953/1500
Epoch 00952: val_loss did not improve
3s - loss: 6.1649e-04 - val_loss: 5.7563e-04
Epoch 954/1500
Epoch 00953: val_loss did not improve
3s - loss: 6.1546e-04 - val_loss: 5.4580e-04
Epoch 955/1500
Epoch 00954: val_loss did not improve
3s - loss: 6.2429e-04 - val_loss: 5.3976e-04
Epoch 956/1500
Epoch 00955: val_loss did not improve
3s - loss: 6.0570e-04 - val_loss: 5.5755e-04
Epoch 957/1500
Epoch 00956: val_loss did not improve
3s - loss: 5.9793e-04 - val_loss: 5.3845e-04
Epoch 958/1500
Epoch 00957: val_loss did not improve
3s - loss: 5.9650e-04 - val_loss: 5.6234e-04
Epoch 959/1500
Epoch 00958: val_loss did not improve
3s - loss: 6.0739e-04 - val_loss: 5.5105e-04
Epoch 960/1500
Epoch 00959: val_loss did not improve
3s - loss: 6.4014e-04 - val_loss: 5.4939e-04
Epoch 961/1500
Epoch 00960: val_loss did not improve
13s - loss: 6.3685e-04 - val_loss: 5.6889e-04
Epoch 962/1500
Epoch 00961: val_loss did not improve
4s - loss: 6.2667e-04 - val_loss: 5.5204e-04
Epoch 963/1500
Epoch 00962: val_loss did not improve
3s - loss: 6.1876e-04 - val_loss: 5.4089e-04
Epoch 964/1500
Epoch 00963: val_loss did not improve
3s - loss: 6.1612e-04 - val_loss: 5.9078e-04
Epoch 965/1500
Epoch 00964: val_loss did not improve
3s - loss: 6.2623e-04 - val_loss: 5.9798e-04
Epoch 966/1500
Epoch 00965: val_loss did not improve
3s - loss: 6.2759e-04 - val_loss: 5.7120e-04
Epoch 967/1500
Epoch 00966: val_loss did not improve
4s - loss: 6.1674e-04 - val_loss: 5.5705e-04
Epoch 968/1500
Epoch 00967: val_loss did not improve
8s - loss: 6.1429e-04 - val_loss: 5.5924e-04
Epoch 969/1500
Epoch 00968: val_loss did not improve
3s - loss: 5.9332e-04 - val_loss: 5.6838e-04
Epoch 970/1500
Epoch 00969: val_loss did not improve
3s - loss: 6.2325e-04 - val_loss: 5.4142e-04
Epoch 971/1500
Epoch 00970: val_loss did not improve
3s - loss: 6.0045e-04 - val_loss: 5.5731e-04
Epoch 972/1500
Epoch 00971: val_loss did not improve
3s - loss: 6.2058e-04 - val_loss: 5.4607e-04
Epoch 973/1500
Epoch 00972: val_loss did not improve
3s - loss: 6.2316e-04 - val_loss: 5.7547e-04
Epoch 974/1500
Epoch 00973: val_loss did not improve
3s - loss: 6.2286e-04 - val_loss: 5.5163e-04
Epoch 975/1500
Epoch 00974: val_loss did not improve
3s - loss: 6.0816e-04 - val_loss: 5.5560e-04
Epoch 976/1500
Epoch 00975: val_loss did not improve
3s - loss: 6.2564e-04 - val_loss: 5.7776e-04
Epoch 977/1500
Epoch 00976: val_loss did not improve
5s - loss: 6.1418e-04 - val_loss: 5.7978e-04
Epoch 978/1500
Epoch 00977: val_loss did not improve
16s - loss: 6.1275e-04 - val_loss: 5.7792e-04
Epoch 979/1500
Epoch 00978: val_loss did not improve
4s - loss: 6.3603e-04 - val_loss: 5.4922e-04
Epoch 980/1500
Epoch 00979: val_loss did not improve
4s - loss: 5.9860e-04 - val_loss: 5.7035e-04
Epoch 981/1500
Epoch 00980: val_loss did not improve
4s - loss: 6.1152e-04 - val_loss: 5.5880e-04
Epoch 982/1500
Epoch 00981: val_loss did not improve
4s - loss: 6.1616e-04 - val_loss: 5.7515e-04
Epoch 983/1500
Epoch 00982: val_loss did not improve
4s - loss: 6.1854e-04 - val_loss: 5.5041e-04
Epoch 984/1500
Epoch 00983: val_loss did not improve
4s - loss: 6.1349e-04 - val_loss: 5.5912e-04
Epoch 985/1500
Epoch 00984: val_loss did not improve
6s - loss: 6.2088e-04 - val_loss: 5.6127e-04
Epoch 986/1500
Epoch 00985: val_loss did not improve
4s - loss: 6.1738e-04 - val_loss: 5.5495e-04
Epoch 987/1500
Epoch 00986: val_loss did not improve
3s - loss: 6.1600e-04 - val_loss: 5.5179e-04
Epoch 988/1500
Epoch 00987: val_loss did not improve
3s - loss: 6.2157e-04 - val_loss: 5.6228e-04
Epoch 989/1500
Epoch 00988: val_loss did not improve
3s - loss: 6.1683e-04 - val_loss: 5.4926e-04
Epoch 990/1500
Epoch 00989: val_loss did not improve
3s - loss: 6.2489e-04 - val_loss: 5.6051e-04
Epoch 991/1500
Epoch 00990: val_loss did not improve
3s - loss: 6.3479e-04 - val_loss: 5.5849e-04
Epoch 992/1500
Epoch 00991: val_loss did not improve
3s - loss: 6.0118e-04 - val_loss: 5.7431e-04
Epoch 993/1500
Epoch 00992: val_loss did not improve
8s - loss: 6.2157e-04 - val_loss: 5.5632e-04
Epoch 994/1500
Epoch 00993: val_loss did not improve
7s - loss: 6.0586e-04 - val_loss: 6.0160e-04
Epoch 995/1500
Epoch 00994: val_loss did not improve
8s - loss: 5.9206e-04 - val_loss: 6.0422e-04
Epoch 996/1500
Epoch 00995: val_loss did not improve
7s - loss: 6.2242e-04 - val_loss: 5.7824e-04
Epoch 997/1500
Epoch 00996: val_loss did not improve
4s - loss: 6.3164e-04 - val_loss: 5.4112e-04
Epoch 998/1500
Epoch 00997: val_loss did not improve
3s - loss: 6.0611e-04 - val_loss: 5.7775e-04
Epoch 999/1500
Epoch 00998: val_loss did not improve
3s - loss: 5.9997e-04 - val_loss: 5.6986e-04
Epoch 1000/1500
Epoch 00999: val_loss did not improve
3s - loss: 6.0802e-04 - val_loss: 5.7588e-04
Epoch 1001/1500
Epoch 01000: val_loss did not improve
3s - loss: 6.1763e-04 - val_loss: 5.5807e-04
Epoch 1002/1500
Epoch 01001: val_loss did not improve
3s - loss: 6.0577e-04 - val_loss: 5.6056e-04
Epoch 1003/1500
Epoch 01002: val_loss did not improve
3s - loss: 6.0811e-04 - val_loss: 5.4720e-04
Epoch 1004/1500
Epoch 01003: val_loss did not improve
3s - loss: 6.4361e-04 - val_loss: 5.6450e-04
Epoch 1005/1500
Epoch 01004: val_loss did not improve
3s - loss: 6.2154e-04 - val_loss: 5.7511e-04
Epoch 1006/1500
Epoch 01005: val_loss did not improve
3s - loss: 6.1887e-04 - val_loss: 5.5267e-04
Epoch 1007/1500
Epoch 01006: val_loss did not improve
3s - loss: 6.1155e-04 - val_loss: 5.6545e-04
Epoch 1008/1500
Epoch 01007: val_loss did not improve
3s - loss: 6.0303e-04 - val_loss: 5.7694e-04
Epoch 1009/1500
Epoch 01008: val_loss did not improve
3s - loss: 6.2330e-04 - val_loss: 5.4813e-04
Epoch 1010/1500
Epoch 01009: val_loss did not improve
3s - loss: 6.0759e-04 - val_loss: 5.9109e-04
Epoch 1011/1500
Epoch 01010: val_loss did not improve
3s - loss: 6.1017e-04 - val_loss: 5.7362e-04
Epoch 1012/1500
Epoch 01011: val_loss did not improve
3s - loss: 6.1820e-04 - val_loss: 5.8664e-04
Epoch 1013/1500
Epoch 01012: val_loss did not improve
3s - loss: 6.0767e-04 - val_loss: 5.6024e-04
Epoch 1014/1500
Epoch 01013: val_loss did not improve
3s - loss: 5.9684e-04 - val_loss: 5.7105e-04
Epoch 1015/1500
Epoch 01014: val_loss did not improve
3s - loss: 6.1058e-04 - val_loss: 5.6025e-04
Epoch 1016/1500
Epoch 01015: val_loss did not improve
7s - loss: 6.0743e-04 - val_loss: 5.4836e-04
Epoch 1017/1500
Epoch 01016: val_loss did not improve
8s - loss: 6.1774e-04 - val_loss: 5.7888e-04
Epoch 1018/1500
Epoch 01017: val_loss did not improve
7s - loss: 6.1592e-04 - val_loss: 5.7652e-04
Epoch 1019/1500
Epoch 01018: val_loss did not improve
7s - loss: 6.1441e-04 - val_loss: 5.6008e-04
Epoch 1020/1500
Epoch 01019: val_loss did not improve
7s - loss: 6.0365e-04 - val_loss: 5.6025e-04
Epoch 1021/1500
Epoch 01020: val_loss did not improve
7s - loss: 5.9874e-04 - val_loss: 5.6006e-04
Epoch 1022/1500
Epoch 01021: val_loss did not improve
7s - loss: 6.2343e-04 - val_loss: 5.6019e-04
Epoch 1023/1500
Epoch 01022: val_loss did not improve
4s - loss: 6.1385e-04 - val_loss: 5.5995e-04
Epoch 1024/1500
Epoch 01023: val_loss did not improve
4s - loss: 6.1980e-04 - val_loss: 5.4340e-04
Epoch 1025/1500
Epoch 01024: val_loss did not improve
3s - loss: 5.9648e-04 - val_loss: 5.3976e-04
Epoch 1026/1500
Epoch 01025: val_loss did not improve
3s - loss: 6.1735e-04 - val_loss: 5.5445e-04
Epoch 1027/1500
Epoch 01026: val_loss did not improve
4s - loss: 6.0241e-04 - val_loss: 5.5515e-04
Epoch 1028/1500
Epoch 01027: val_loss did not improve
15s - loss: 5.9822e-04 - val_loss: 5.6232e-04
Epoch 1029/1500
Epoch 01028: val_loss did not improve
4s - loss: 6.0531e-04 - val_loss: 5.6875e-04
Epoch 1030/1500
Epoch 01029: val_loss did not improve
3s - loss: 6.1721e-04 - val_loss: 5.9381e-04
Epoch 1031/1500
Epoch 01030: val_loss did not improve
3s - loss: 6.1168e-04 - val_loss: 5.4734e-04
Epoch 1032/1500
Epoch 01031: val_loss improved from 0.00053 to 0.00053, saving model to my_model.h5
3s - loss: 6.3327e-04 - val_loss: 5.3403e-04
Epoch 1033/1500
Epoch 01032: val_loss did not improve
3s - loss: 6.1232e-04 - val_loss: 5.5327e-04
Epoch 1034/1500
Epoch 01033: val_loss did not improve
3s - loss: 6.0038e-04 - val_loss: 5.7351e-04
Epoch 1035/1500
Epoch 01034: val_loss did not improve
3s - loss: 6.0753e-04 - val_loss: 5.7033e-04
Epoch 1036/1500
Epoch 01035: val_loss did not improve
3s - loss: 6.0394e-04 - val_loss: 5.7663e-04
Epoch 1037/1500
Epoch 01036: val_loss did not improve
3s - loss: 6.1094e-04 - val_loss: 5.6172e-04
Epoch 1038/1500
Epoch 01037: val_loss did not improve
3s - loss: 6.2575e-04 - val_loss: 5.7699e-04
Epoch 1039/1500
Epoch 01038: val_loss did not improve
3s - loss: 5.9659e-04 - val_loss: 5.5660e-04
Epoch 1040/1500
Epoch 01039: val_loss did not improve
3s - loss: 6.0302e-04 - val_loss: 5.6392e-04
Epoch 1041/1500
Epoch 01040: val_loss did not improve
3s - loss: 6.0878e-04 - val_loss: 5.5640e-04
Epoch 1042/1500
Epoch 01041: val_loss did not improve
3s - loss: 6.1411e-04 - val_loss: 5.6787e-04
Epoch 1043/1500
Epoch 01042: val_loss did not improve
3s - loss: 5.9647e-04 - val_loss: 5.5939e-04
Epoch 1044/1500
Epoch 01043: val_loss did not improve
3s - loss: 5.9574e-04 - val_loss: 5.7863e-04
Epoch 1045/1500
Epoch 01044: val_loss did not improve
3s - loss: 5.9000e-04 - val_loss: 5.8402e-04
Epoch 1046/1500
Epoch 01045: val_loss did not improve
3s - loss: 6.0833e-04 - val_loss: 5.7011e-04
Epoch 1047/1500
Epoch 01046: val_loss did not improve
3s - loss: 6.2059e-04 - val_loss: 5.4643e-04
Epoch 1048/1500
Epoch 01047: val_loss did not improve
3s - loss: 6.0376e-04 - val_loss: 5.5985e-04
Epoch 1049/1500
Epoch 01048: val_loss did not improve
3s - loss: 5.9562e-04 - val_loss: 5.6063e-04
Epoch 1050/1500
Epoch 01049: val_loss did not improve
3s - loss: 6.0123e-04 - val_loss: 5.7483e-04
Epoch 1051/1500
Epoch 01050: val_loss did not improve
3s - loss: 6.0289e-04 - val_loss: 5.7341e-04
Epoch 1052/1500
Epoch 01051: val_loss did not improve
3s - loss: 5.9599e-04 - val_loss: 5.6635e-04
Epoch 1053/1500
Epoch 01052: val_loss did not improve
3s - loss: 6.0665e-04 - val_loss: 5.7997e-04
Epoch 1054/1500
Epoch 01053: val_loss did not improve
3s - loss: 5.8955e-04 - val_loss: 5.6725e-04
Epoch 1055/1500
Epoch 01054: val_loss did not improve
3s - loss: 5.9503e-04 - val_loss: 5.8408e-04
Epoch 1056/1500
Epoch 01055: val_loss did not improve
3s - loss: 6.0754e-04 - val_loss: 5.7247e-04
Epoch 1057/1500
Epoch 01056: val_loss did not improve
3s - loss: 5.8670e-04 - val_loss: 5.5734e-04
Epoch 1058/1500
Epoch 01057: val_loss did not improve
3s - loss: 6.0911e-04 - val_loss: 5.5337e-04
Epoch 1059/1500
Epoch 01058: val_loss did not improve
3s - loss: 6.1998e-04 - val_loss: 5.4611e-04
Epoch 1060/1500
Epoch 01059: val_loss did not improve
3s - loss: 5.6863e-04 - val_loss: 5.3951e-04
Epoch 1061/1500
Epoch 01060: val_loss did not improve
3s - loss: 6.0229e-04 - val_loss: 5.5320e-04
Epoch 1062/1500
Epoch 01061: val_loss did not improve
3s - loss: 6.0259e-04 - val_loss: 5.7893e-04
Epoch 1063/1500
Epoch 01062: val_loss did not improve
3s - loss: 5.9271e-04 - val_loss: 5.9247e-04
Epoch 1064/1500
Epoch 01063: val_loss did not improve
3s - loss: 6.1367e-04 - val_loss: 5.7016e-04
Epoch 1065/1500
Epoch 01064: val_loss did not improve
3s - loss: 6.1054e-04 - val_loss: 5.5722e-04
Epoch 1066/1500
Epoch 01065: val_loss did not improve
3s - loss: 5.9477e-04 - val_loss: 5.7524e-04
Epoch 1067/1500
Epoch 01066: val_loss did not improve
3s - loss: 6.0389e-04 - val_loss: 5.5127e-04
Epoch 1068/1500
Epoch 01067: val_loss did not improve
3s - loss: 5.9038e-04 - val_loss: 5.6364e-04
Epoch 1069/1500
Epoch 01068: val_loss did not improve
4s - loss: 6.1162e-04 - val_loss: 5.8662e-04
Epoch 1070/1500
Epoch 01069: val_loss did not improve
4s - loss: 6.1224e-04 - val_loss: 5.5827e-04
Epoch 1071/1500
Epoch 01070: val_loss did not improve
3s - loss: 5.9642e-04 - val_loss: 5.5625e-04
Epoch 1072/1500
Epoch 01071: val_loss did not improve
3s - loss: 5.7779e-04 - val_loss: 5.5029e-04
Epoch 1073/1500
Epoch 01072: val_loss did not improve
3s - loss: 6.1746e-04 - val_loss: 5.8010e-04
Epoch 1074/1500
Epoch 01073: val_loss did not improve
3s - loss: 6.0015e-04 - val_loss: 5.8607e-04
Epoch 1075/1500
Epoch 01074: val_loss did not improve
3s - loss: 6.2759e-04 - val_loss: 5.6741e-04
Epoch 1076/1500
Epoch 01075: val_loss did not improve
3s - loss: 6.1254e-04 - val_loss: 5.6569e-04
Epoch 1077/1500
Epoch 01076: val_loss did not improve
3s - loss: 6.1181e-04 - val_loss: 5.4803e-04
Epoch 1078/1500
Epoch 01077: val_loss did not improve
3s - loss: 5.9679e-04 - val_loss: 5.8055e-04
Epoch 1079/1500
Epoch 01078: val_loss did not improve
3s - loss: 6.0619e-04 - val_loss: 5.7457e-04
Epoch 1080/1500
Epoch 01079: val_loss did not improve
3s - loss: 5.9802e-04 - val_loss: 5.6819e-04
Epoch 1081/1500
Epoch 01080: val_loss did not improve
3s - loss: 6.0394e-04 - val_loss: 5.7375e-04
Epoch 1082/1500
Epoch 01081: val_loss did not improve
3s - loss: 6.2519e-04 - val_loss: 5.7375e-04
Epoch 1083/1500
Epoch 01082: val_loss did not improve
3s - loss: 6.1538e-04 - val_loss: 5.9432e-04
Epoch 1084/1500
Epoch 01083: val_loss did not improve
3s - loss: 6.2536e-04 - val_loss: 5.8360e-04
Epoch 1085/1500
Epoch 01084: val_loss did not improve
3s - loss: 6.0730e-04 - val_loss: 5.5644e-04
Epoch 1086/1500
Epoch 01085: val_loss did not improve
3s - loss: 5.9901e-04 - val_loss: 5.5708e-04
Epoch 1087/1500
Epoch 01086: val_loss did not improve
3s - loss: 5.9834e-04 - val_loss: 5.6028e-04
Epoch 1088/1500
Epoch 01087: val_loss did not improve
3s - loss: 5.8886e-04 - val_loss: 5.5177e-04
Epoch 1089/1500
Epoch 01088: val_loss did not improve
3s - loss: 5.9422e-04 - val_loss: 5.7376e-04
Epoch 1090/1500
Epoch 01089: val_loss did not improve
3s - loss: 5.9735e-04 - val_loss: 5.8042e-04
Epoch 1091/1500
Epoch 01090: val_loss did not improve
3s - loss: 5.9777e-04 - val_loss: 5.6354e-04
Epoch 1092/1500
Epoch 01091: val_loss did not improve
3s - loss: 5.8360e-04 - val_loss: 5.7689e-04
Epoch 1093/1500
Epoch 01092: val_loss did not improve
3s - loss: 6.0655e-04 - val_loss: 5.6350e-04
Epoch 1094/1500
Epoch 01093: val_loss did not improve
3s - loss: 5.9391e-04 - val_loss: 5.6334e-04
Epoch 1095/1500
Epoch 01094: val_loss did not improve
3s - loss: 6.0343e-04 - val_loss: 5.8212e-04
Epoch 1096/1500
Epoch 01095: val_loss did not improve
3s - loss: 5.8785e-04 - val_loss: 5.8897e-04
Epoch 1097/1500
Epoch 01096: val_loss did not improve
3s - loss: 5.9285e-04 - val_loss: 5.5502e-04
Epoch 1098/1500
Epoch 01097: val_loss did not improve
3s - loss: 6.0021e-04 - val_loss: 5.6275e-04
Epoch 1099/1500
Epoch 01098: val_loss did not improve
3s - loss: 6.0553e-04 - val_loss: 5.4830e-04
Epoch 1100/1500
Epoch 01099: val_loss did not improve
3s - loss: 5.9607e-04 - val_loss: 5.5584e-04
Epoch 1101/1500
Epoch 01100: val_loss did not improve
3s - loss: 5.9142e-04 - val_loss: 5.8052e-04
Epoch 1102/1500
Epoch 01101: val_loss did not improve
3s - loss: 6.0046e-04 - val_loss: 5.5104e-04
Epoch 1103/1500
Epoch 01102: val_loss did not improve
3s - loss: 5.7390e-04 - val_loss: 5.6635e-04
Epoch 1104/1500
Epoch 01103: val_loss did not improve
3s - loss: 5.9800e-04 - val_loss: 5.4683e-04
Epoch 1105/1500
Epoch 01104: val_loss did not improve
3s - loss: 6.0885e-04 - val_loss: 5.6921e-04
Epoch 1106/1500
Epoch 01105: val_loss did not improve
3s - loss: 5.8527e-04 - val_loss: 5.8623e-04
Epoch 1107/1500
Epoch 01106: val_loss did not improve
3s - loss: 5.9770e-04 - val_loss: 5.4578e-04
Epoch 1108/1500
Epoch 01107: val_loss did not improve
3s - loss: 6.2128e-04 - val_loss: 5.6046e-04
Epoch 1109/1500
Epoch 01108: val_loss did not improve
3s - loss: 6.1011e-04 - val_loss: 5.6076e-04
Epoch 1110/1500
Epoch 01109: val_loss did not improve
3s - loss: 6.0792e-04 - val_loss: 5.8447e-04
Epoch 1111/1500
Epoch 01110: val_loss did not improve
3s - loss: 6.1142e-04 - val_loss: 5.5709e-04
Epoch 1112/1500
Epoch 01111: val_loss did not improve
3s - loss: 6.3511e-04 - val_loss: 5.5501e-04
Epoch 1113/1500
Epoch 01112: val_loss did not improve
3s - loss: 6.0351e-04 - val_loss: 5.6119e-04
Epoch 1114/1500
Epoch 01113: val_loss did not improve
3s - loss: 5.9472e-04 - val_loss: 5.7120e-04
Epoch 1115/1500
Epoch 01114: val_loss did not improve
3s - loss: 5.9715e-04 - val_loss: 5.6964e-04
Epoch 1116/1500
Epoch 01115: val_loss did not improve
3s - loss: 6.3209e-04 - val_loss: 5.7437e-04
Epoch 1117/1500
Epoch 01116: val_loss did not improve
3s - loss: 6.0448e-04 - val_loss: 5.7097e-04
Epoch 1118/1500
Epoch 01117: val_loss did not improve
3s - loss: 6.0334e-04 - val_loss: 5.6242e-04
Epoch 1119/1500
Epoch 01118: val_loss did not improve
3s - loss: 5.7966e-04 - val_loss: 5.6436e-04
Epoch 1120/1500
Epoch 01119: val_loss did not improve
3s - loss: 5.9815e-04 - val_loss: 5.4579e-04
Epoch 1121/1500
Epoch 01120: val_loss did not improve
3s - loss: 6.2179e-04 - val_loss: 5.7882e-04
Epoch 1122/1500
Epoch 01121: val_loss did not improve
3s - loss: 6.0500e-04 - val_loss: 5.5900e-04
Epoch 1123/1500
Epoch 01122: val_loss did not improve
3s - loss: 5.9694e-04 - val_loss: 5.6720e-04
Epoch 1124/1500
Epoch 01123: val_loss did not improve
3s - loss: 6.1131e-04 - val_loss: 5.5613e-04
Epoch 1125/1500
Epoch 01124: val_loss did not improve
3s - loss: 5.7444e-04 - val_loss: 5.6345e-04
Epoch 1126/1500
Epoch 01125: val_loss did not improve
3s - loss: 5.9653e-04 - val_loss: 5.8722e-04
Epoch 1127/1500
Epoch 01126: val_loss did not improve
3s - loss: 5.9736e-04 - val_loss: 5.5757e-04
Epoch 1128/1500
Epoch 01127: val_loss did not improve
3s - loss: 5.9053e-04 - val_loss: 5.4173e-04
Epoch 1129/1500
Epoch 01128: val_loss did not improve
3s - loss: 5.7794e-04 - val_loss: 5.6704e-04
Epoch 1130/1500
Epoch 01129: val_loss did not improve
3s - loss: 5.8083e-04 - val_loss: 5.7949e-04
Epoch 1131/1500
Epoch 01130: val_loss did not improve
3s - loss: 5.7414e-04 - val_loss: 5.6679e-04
Epoch 1132/1500
Epoch 01131: val_loss did not improve
3s - loss: 5.8731e-04 - val_loss: 5.4683e-04
Epoch 1133/1500
Epoch 01132: val_loss did not improve
3s - loss: 5.8411e-04 - val_loss: 5.6912e-04
Epoch 1134/1500
Epoch 01133: val_loss did not improve
3s - loss: 5.8013e-04 - val_loss: 5.6221e-04
Epoch 1135/1500
Epoch 01134: val_loss did not improve
3s - loss: 5.9316e-04 - val_loss: 5.6947e-04
Epoch 1136/1500
Epoch 01135: val_loss did not improve
3s - loss: 5.9500e-04 - val_loss: 5.8716e-04
Epoch 1137/1500
Epoch 01136: val_loss did not improve
3s - loss: 5.8132e-04 - val_loss: 5.5641e-04
Epoch 1138/1500
Epoch 01137: val_loss did not improve
3s - loss: 5.7834e-04 - val_loss: 5.4836e-04
Epoch 1139/1500
Epoch 01138: val_loss did not improve
3s - loss: 6.0135e-04 - val_loss: 5.5795e-04
Epoch 1140/1500
Epoch 01139: val_loss did not improve
3s - loss: 5.9766e-04 - val_loss: 5.5645e-04
Epoch 1141/1500
Epoch 01140: val_loss did not improve
3s - loss: 5.9228e-04 - val_loss: 5.5337e-04
Epoch 1142/1500
Epoch 01141: val_loss did not improve
3s - loss: 5.8273e-04 - val_loss: 5.5066e-04
Epoch 1143/1500
Epoch 01142: val_loss did not improve
3s - loss: 5.9824e-04 - val_loss: 5.9091e-04
Epoch 1144/1500
Epoch 01143: val_loss did not improve
3s - loss: 5.9554e-04 - val_loss: 6.0168e-04
Epoch 1145/1500
Epoch 01144: val_loss did not improve
3s - loss: 6.1581e-04 - val_loss: 5.6771e-04
Epoch 1146/1500
Epoch 01145: val_loss did not improve
3s - loss: 6.1103e-04 - val_loss: 6.0834e-04
Epoch 1147/1500
Epoch 01146: val_loss did not improve
3s - loss: 6.0457e-04 - val_loss: 5.7247e-04
Epoch 1148/1500
Epoch 01147: val_loss did not improve
3s - loss: 5.9624e-04 - val_loss: 5.5361e-04
Epoch 1149/1500
Epoch 01148: val_loss did not improve
3s - loss: 5.8055e-04 - val_loss: 5.6792e-04
Epoch 1150/1500
Epoch 01149: val_loss did not improve
3s - loss: 6.0961e-04 - val_loss: 5.6476e-04
Epoch 1151/1500
Epoch 01150: val_loss did not improve
3s - loss: 6.0202e-04 - val_loss: 5.8772e-04
Epoch 1152/1500
Epoch 01151: val_loss did not improve
3s - loss: 5.9657e-04 - val_loss: 5.7620e-04
Epoch 1153/1500
Epoch 01152: val_loss did not improve
3s - loss: 5.9730e-04 - val_loss: 5.6896e-04
Epoch 1154/1500
Epoch 01153: val_loss did not improve
3s - loss: 6.1345e-04 - val_loss: 5.6756e-04
Epoch 1155/1500
Epoch 01154: val_loss did not improve
3s - loss: 5.9879e-04 - val_loss: 5.6669e-04
Epoch 1156/1500
Epoch 01155: val_loss did not improve
3s - loss: 5.7531e-04 - val_loss: 5.6521e-04
Epoch 1157/1500
Epoch 01156: val_loss did not improve
3s - loss: 5.9077e-04 - val_loss: 5.6546e-04
Epoch 1158/1500
Epoch 01157: val_loss did not improve
3s - loss: 6.2205e-04 - val_loss: 5.7615e-04
Epoch 1159/1500
Epoch 01158: val_loss did not improve
3s - loss: 5.6377e-04 - val_loss: 5.4927e-04
Epoch 1160/1500
Epoch 01159: val_loss did not improve
3s - loss: 5.8617e-04 - val_loss: 5.8299e-04
Epoch 1161/1500
Epoch 01160: val_loss did not improve
3s - loss: 6.0355e-04 - val_loss: 5.5078e-04
Epoch 1162/1500
Epoch 01161: val_loss did not improve
3s - loss: 6.1006e-04 - val_loss: 5.5024e-04
Epoch 1163/1500
Epoch 01162: val_loss did not improve
3s - loss: 5.9844e-04 - val_loss: 5.6161e-04
Epoch 1164/1500
Epoch 01163: val_loss did not improve
3s - loss: 5.8078e-04 - val_loss: 5.5857e-04
Epoch 1165/1500
Epoch 01164: val_loss did not improve
3s - loss: 6.0245e-04 - val_loss: 5.9340e-04
Epoch 1166/1500
Epoch 01165: val_loss did not improve
3s - loss: 6.0116e-04 - val_loss: 6.0397e-04
Epoch 1167/1500
Epoch 01166: val_loss did not improve
3s - loss: 5.8886e-04 - val_loss: 5.6075e-04
Epoch 1168/1500
Epoch 01167: val_loss did not improve
3s - loss: 5.9923e-04 - val_loss: 5.9911e-04
Epoch 1169/1500
Epoch 01168: val_loss did not improve
3s - loss: 5.8956e-04 - val_loss: 5.4870e-04
Epoch 1170/1500
Epoch 01169: val_loss did not improve
3s - loss: 5.9177e-04 - val_loss: 5.4951e-04
Epoch 1171/1500
Epoch 01170: val_loss did not improve
3s - loss: 5.8609e-04 - val_loss: 5.8497e-04
Epoch 1172/1500
Epoch 01171: val_loss did not improve
3s - loss: 6.0019e-04 - val_loss: 5.7293e-04
Epoch 1173/1500
Epoch 01172: val_loss did not improve
3s - loss: 5.8442e-04 - val_loss: 5.7220e-04
Epoch 1174/1500
Epoch 01173: val_loss did not improve
3s - loss: 5.9613e-04 - val_loss: 5.6730e-04
Epoch 1175/1500
Epoch 01174: val_loss did not improve
3s - loss: 6.0301e-04 - val_loss: 5.6923e-04
Epoch 1176/1500
Epoch 01175: val_loss did not improve
3s - loss: 5.9508e-04 - val_loss: 5.6387e-04
Epoch 1177/1500
Epoch 01176: val_loss did not improve
3s - loss: 5.8885e-04 - val_loss: 6.0726e-04
Epoch 1178/1500
Epoch 01177: val_loss did not improve
3s - loss: 6.0819e-04 - val_loss: 5.7239e-04
Epoch 1179/1500
Epoch 01178: val_loss did not improve
3s - loss: 5.6543e-04 - val_loss: 5.7213e-04
Epoch 1180/1500
Epoch 01179: val_loss did not improve
3s - loss: 5.8264e-04 - val_loss: 5.8258e-04
Epoch 1181/1500
Epoch 01180: val_loss did not improve
3s - loss: 5.8119e-04 - val_loss: 5.7762e-04
Epoch 1182/1500
Epoch 01181: val_loss did not improve
3s - loss: 5.7558e-04 - val_loss: 5.7631e-04
Epoch 1183/1500
Epoch 01182: val_loss did not improve
3s - loss: 5.8141e-04 - val_loss: 5.5720e-04
Epoch 1184/1500
Epoch 01183: val_loss did not improve
3s - loss: 5.9323e-04 - val_loss: 5.4742e-04
Epoch 1185/1500
Epoch 01184: val_loss did not improve
3s - loss: 5.8020e-04 - val_loss: 5.4972e-04
Epoch 1186/1500
Epoch 01185: val_loss did not improve
3s - loss: 5.7901e-04 - val_loss: 5.4910e-04
Epoch 1187/1500
Epoch 01186: val_loss did not improve
3s - loss: 5.9593e-04 - val_loss: 5.5543e-04
Epoch 1188/1500
Epoch 01187: val_loss did not improve
3s - loss: 5.8859e-04 - val_loss: 5.8184e-04
Epoch 1189/1500
Epoch 01188: val_loss did not improve
3s - loss: 5.9036e-04 - val_loss: 5.5527e-04
Epoch 1190/1500
Epoch 01189: val_loss did not improve
3s - loss: 5.9174e-04 - val_loss: 5.8669e-04
Epoch 1191/1500
Epoch 01190: val_loss did not improve
3s - loss: 5.9519e-04 - val_loss: 5.7411e-04
Epoch 1192/1500
Epoch 01191: val_loss did not improve
3s - loss: 6.1425e-04 - val_loss: 5.5595e-04
Epoch 1193/1500
Epoch 01192: val_loss did not improve
3s - loss: 6.0518e-04 - val_loss: 5.6305e-04
Epoch 1194/1500
Epoch 01193: val_loss did not improve
3s - loss: 5.8616e-04 - val_loss: 5.8894e-04
Epoch 1195/1500
Epoch 01194: val_loss did not improve
3s - loss: 6.0175e-04 - val_loss: 5.9488e-04
Epoch 1196/1500
Epoch 01195: val_loss did not improve
3s - loss: 5.9294e-04 - val_loss: 5.5101e-04
Epoch 1197/1500
Epoch 01196: val_loss did not improve
3s - loss: 6.0125e-04 - val_loss: 5.6321e-04
Epoch 1198/1500
Epoch 01197: val_loss did not improve
3s - loss: 5.9402e-04 - val_loss: 5.6725e-04
Epoch 1199/1500
Epoch 01198: val_loss did not improve
3s - loss: 5.9449e-04 - val_loss: 5.6817e-04
Epoch 1200/1500
Epoch 01199: val_loss did not improve
3s - loss: 6.0716e-04 - val_loss: 5.7856e-04
Epoch 1201/1500
Epoch 01200: val_loss did not improve
3s - loss: 6.1333e-04 - val_loss: 5.6481e-04
Epoch 1202/1500
Epoch 01201: val_loss did not improve
3s - loss: 6.0659e-04 - val_loss: 5.5109e-04
Epoch 1203/1500
Epoch 01202: val_loss did not improve
3s - loss: 5.9643e-04 - val_loss: 5.5056e-04
Epoch 1204/1500
Epoch 01203: val_loss did not improve
3s - loss: 6.1733e-04 - val_loss: 5.5967e-04
Epoch 1205/1500
Epoch 01204: val_loss did not improve
3s - loss: 5.9026e-04 - val_loss: 5.5181e-04
Epoch 1206/1500
Epoch 01205: val_loss did not improve
3s - loss: 5.9236e-04 - val_loss: 5.5485e-04
Epoch 1207/1500
Epoch 01206: val_loss did not improve
3s - loss: 5.9079e-04 - val_loss: 5.6867e-04
Epoch 1208/1500
Epoch 01207: val_loss did not improve
3s - loss: 5.8761e-04 - val_loss: 5.6712e-04
Epoch 1209/1500
Epoch 01208: val_loss did not improve
3s - loss: 5.8159e-04 - val_loss: 5.9059e-04
Epoch 1210/1500
Epoch 01209: val_loss did not improve
3s - loss: 5.7811e-04 - val_loss: 5.9739e-04
Epoch 1211/1500
Epoch 01210: val_loss did not improve
3s - loss: 5.9503e-04 - val_loss: 5.6820e-04
Epoch 1212/1500
Epoch 01211: val_loss did not improve
3s - loss: 6.0045e-04 - val_loss: 5.5642e-04
Epoch 1213/1500
Epoch 01212: val_loss did not improve
3s - loss: 5.9097e-04 - val_loss: 5.6436e-04
Epoch 1214/1500
Epoch 01213: val_loss did not improve
3s - loss: 5.7779e-04 - val_loss: 5.6867e-04
Epoch 1215/1500
Epoch 01214: val_loss did not improve
5s - loss: 6.0941e-04 - val_loss: 5.7622e-04
Epoch 1216/1500
Epoch 01215: val_loss did not improve
5s - loss: 5.9421e-04 - val_loss: 5.5769e-04
Epoch 1217/1500
Epoch 01216: val_loss did not improve
5s - loss: 5.9088e-04 - val_loss: 5.6417e-04
Epoch 1218/1500
Epoch 01217: val_loss did not improve
5s - loss: 5.8618e-04 - val_loss: 5.7217e-04
Epoch 1219/1500
Epoch 01218: val_loss did not improve
3s - loss: 5.8785e-04 - val_loss: 5.6981e-04
Epoch 1220/1500
Epoch 01219: val_loss did not improve
3s - loss: 5.9905e-04 - val_loss: 5.6766e-04
Epoch 1221/1500
Epoch 01220: val_loss did not improve
3s - loss: 5.8787e-04 - val_loss: 5.8551e-04
Epoch 1222/1500
Epoch 01221: val_loss did not improve
3s - loss: 5.8436e-04 - val_loss: 5.5352e-04
Epoch 1223/1500
Epoch 01222: val_loss did not improve
7s - loss: 5.8869e-04 - val_loss: 5.5692e-04
Epoch 1224/1500
Epoch 01223: val_loss did not improve
7s - loss: 5.8869e-04 - val_loss: 5.5702e-04
Epoch 1225/1500
Epoch 01224: val_loss did not improve
7s - loss: 5.7828e-04 - val_loss: 5.9244e-04
Epoch 1226/1500
Epoch 01225: val_loss did not improve
7s - loss: 5.8167e-04 - val_loss: 5.8285e-04
Epoch 1227/1500
Epoch 01226: val_loss did not improve
7s - loss: 6.0268e-04 - val_loss: 5.6962e-04
Epoch 1228/1500
Epoch 01227: val_loss did not improve
6s - loss: 5.8139e-04 - val_loss: 5.7253e-04
Epoch 1229/1500
Epoch 01228: val_loss did not improve
5s - loss: 5.7729e-04 - val_loss: 5.7540e-04
Epoch 1230/1500
Epoch 01229: val_loss did not improve
5s - loss: 5.9110e-04 - val_loss: 5.8059e-04
Epoch 1231/1500
Epoch 01230: val_loss did not improve
6s - loss: 5.9788e-04 - val_loss: 5.7373e-04
Epoch 1232/1500
Epoch 01231: val_loss did not improve
4s - loss: 5.7740e-04 - val_loss: 5.7524e-04
Epoch 1233/1500
Epoch 01232: val_loss did not improve
3s - loss: 5.7310e-04 - val_loss: 5.7153e-04
Epoch 1234/1500
Epoch 01233: val_loss did not improve
3s - loss: 5.6605e-04 - val_loss: 5.8542e-04
Epoch 1235/1500
Epoch 01234: val_loss did not improve
3s - loss: 5.8859e-04 - val_loss: 5.6428e-04
Epoch 1236/1500
Epoch 01235: val_loss did not improve
3s - loss: 5.8792e-04 - val_loss: 5.5004e-04
Epoch 1237/1500
Epoch 01236: val_loss did not improve
3s - loss: 6.0807e-04 - val_loss: 5.4802e-04
Epoch 1238/1500
Epoch 01237: val_loss did not improve
3s - loss: 5.8978e-04 - val_loss: 5.9516e-04
Epoch 1239/1500
Epoch 01238: val_loss did not improve
3s - loss: 5.9034e-04 - val_loss: 6.1586e-04
Epoch 1240/1500
Epoch 01239: val_loss did not improve
3s - loss: 6.0674e-04 - val_loss: 5.5044e-04
Epoch 1241/1500
Epoch 01240: val_loss did not improve
3s - loss: 5.8748e-04 - val_loss: 5.6163e-04
Epoch 1242/1500
Epoch 01241: val_loss did not improve
3s - loss: 5.8322e-04 - val_loss: 5.7448e-04
Epoch 1243/1500
Epoch 01242: val_loss did not improve
3s - loss: 6.0129e-04 - val_loss: 5.9464e-04
Epoch 1244/1500
Epoch 01243: val_loss did not improve
3s - loss: 5.8412e-04 - val_loss: 5.6266e-04
Epoch 1245/1500
Epoch 01244: val_loss did not improve
3s - loss: 5.8785e-04 - val_loss: 5.6766e-04
Epoch 1246/1500
Epoch 01245: val_loss did not improve
3s - loss: 5.8783e-04 - val_loss: 5.6705e-04
Epoch 1247/1500
Epoch 01246: val_loss did not improve
3s - loss: 5.9131e-04 - val_loss: 5.8100e-04
Epoch 1248/1500
Epoch 01247: val_loss did not improve
3s - loss: 5.9546e-04 - val_loss: 5.7371e-04
Epoch 1249/1500
Epoch 01248: val_loss did not improve
3s - loss: 5.7424e-04 - val_loss: 5.7253e-04
Epoch 1250/1500
Epoch 01249: val_loss did not improve
3s - loss: 6.0375e-04 - val_loss: 5.5212e-04
Epoch 1251/1500
Epoch 01250: val_loss did not improve
3s - loss: 5.6680e-04 - val_loss: 5.5625e-04
Epoch 1252/1500
Epoch 01251: val_loss did not improve
3s - loss: 5.8951e-04 - val_loss: 5.7837e-04
Epoch 1253/1500
Epoch 01252: val_loss did not improve
3s - loss: 5.8198e-04 - val_loss: 5.6862e-04
Epoch 1254/1500
Epoch 01253: val_loss did not improve
7s - loss: 5.6467e-04 - val_loss: 5.5815e-04
Epoch 1255/1500
Epoch 01254: val_loss did not improve
5s - loss: 5.9695e-04 - val_loss: 5.4651e-04
Epoch 1256/1500
Epoch 01255: val_loss did not improve
3s - loss: 5.9209e-04 - val_loss: 5.6253e-04
Epoch 1257/1500
Epoch 01256: val_loss did not improve
3s - loss: 5.8537e-04 - val_loss: 5.7134e-04
Epoch 1258/1500
Epoch 01257: val_loss did not improve
3s - loss: 5.7292e-04 - val_loss: 5.6913e-04
Epoch 1259/1500
Epoch 01258: val_loss did not improve
3s - loss: 5.7312e-04 - val_loss: 5.6865e-04
Epoch 1260/1500
Epoch 01259: val_loss did not improve
3s - loss: 6.0269e-04 - val_loss: 5.6131e-04
Epoch 1261/1500
Epoch 01260: val_loss did not improve
3s - loss: 5.8128e-04 - val_loss: 5.5566e-04
Epoch 1262/1500
Epoch 01261: val_loss did not improve
3s - loss: 5.8997e-04 - val_loss: 5.5937e-04
Epoch 1263/1500
Epoch 01262: val_loss did not improve
3s - loss: 5.8547e-04 - val_loss: 5.6663e-04
Epoch 1264/1500
Epoch 01263: val_loss did not improve
3s - loss: 5.8090e-04 - val_loss: 5.8739e-04
Epoch 1265/1500
Epoch 01264: val_loss did not improve
3s - loss: 5.6815e-04 - val_loss: 5.5718e-04
Epoch 1266/1500
Epoch 01265: val_loss did not improve
3s - loss: 5.7777e-04 - val_loss: 5.5665e-04
Epoch 1267/1500
Epoch 01266: val_loss did not improve
3s - loss: 5.8700e-04 - val_loss: 5.8268e-04
Epoch 1268/1500
Epoch 01267: val_loss did not improve
3s - loss: 5.8920e-04 - val_loss: 5.6613e-04
Epoch 1269/1500
Epoch 01268: val_loss did not improve
3s - loss: 5.8706e-04 - val_loss: 5.6592e-04
Epoch 1270/1500
Epoch 01269: val_loss did not improve
3s - loss: 5.8969e-04 - val_loss: 5.7040e-04
Epoch 1271/1500
Epoch 01270: val_loss did not improve
3s - loss: 5.9670e-04 - val_loss: 5.6724e-04
Epoch 1272/1500
Epoch 01271: val_loss did not improve
3s - loss: 5.7021e-04 - val_loss: 5.8335e-04
Epoch 1273/1500
Epoch 01272: val_loss did not improve
4s - loss: 5.9754e-04 - val_loss: 5.5888e-04
Epoch 1274/1500
Epoch 01273: val_loss did not improve
4s - loss: 5.8289e-04 - val_loss: 5.7825e-04
Epoch 1275/1500
Epoch 01274: val_loss did not improve
4s - loss: 5.7800e-04 - val_loss: 5.5325e-04
Epoch 1276/1500
Epoch 01275: val_loss did not improve
4s - loss: 5.9178e-04 - val_loss: 5.6480e-04
Epoch 1277/1500
Epoch 01276: val_loss did not improve
4s - loss: 5.6012e-04 - val_loss: 5.7743e-04
Epoch 1278/1500
Epoch 01277: val_loss did not improve
4s - loss: 6.0368e-04 - val_loss: 5.7095e-04
Epoch 1279/1500
Epoch 01278: val_loss did not improve
4s - loss: 5.9546e-04 - val_loss: 6.2311e-04
Epoch 1280/1500
Epoch 01279: val_loss did not improve
4s - loss: 5.8814e-04 - val_loss: 5.8469e-04
Epoch 1281/1500
Epoch 01280: val_loss did not improve
4s - loss: 5.6992e-04 - val_loss: 5.8291e-04
Epoch 1282/1500
Epoch 01281: val_loss did not improve
4s - loss: 5.8534e-04 - val_loss: 5.9101e-04
Epoch 1283/1500
Epoch 01282: val_loss did not improve
4s - loss: 5.7276e-04 - val_loss: 5.7618e-04
Epoch 1284/1500
Epoch 01283: val_loss did not improve
4s - loss: 5.8538e-04 - val_loss: 5.6379e-04
Epoch 1285/1500
Epoch 01284: val_loss did not improve
3s - loss: 5.7141e-04 - val_loss: 5.7850e-04
Epoch 1286/1500
Epoch 01285: val_loss did not improve
3s - loss: 5.8516e-04 - val_loss: 5.5980e-04
Epoch 1287/1500
Epoch 01286: val_loss did not improve
3s - loss: 6.0650e-04 - val_loss: 5.7034e-04
Epoch 1288/1500
Epoch 01287: val_loss did not improve
3s - loss: 5.9467e-04 - val_loss: 5.7702e-04
Epoch 1289/1500
Epoch 01288: val_loss did not improve
3s - loss: 5.7963e-04 - val_loss: 5.9271e-04
Epoch 1290/1500
Epoch 01289: val_loss did not improve
3s - loss: 5.7985e-04 - val_loss: 5.7779e-04
Epoch 1291/1500
Epoch 01290: val_loss did not improve
3s - loss: 5.9796e-04 - val_loss: 5.6775e-04
Epoch 1292/1500
Epoch 01291: val_loss did not improve
3s - loss: 5.7918e-04 - val_loss: 5.7419e-04
Epoch 1293/1500
Epoch 01292: val_loss did not improve
3s - loss: 5.6540e-04 - val_loss: 5.6887e-04
Epoch 1294/1500
Epoch 01293: val_loss did not improve
3s - loss: 5.8479e-04 - val_loss: 5.9840e-04
Epoch 1295/1500
Epoch 01294: val_loss did not improve
3s - loss: 5.8423e-04 - val_loss: 5.7375e-04
Epoch 1296/1500
Epoch 01295: val_loss did not improve
3s - loss: 5.9679e-04 - val_loss: 5.4456e-04
Epoch 1297/1500
Epoch 01296: val_loss did not improve
3s - loss: 5.9815e-04 - val_loss: 5.6013e-04
Epoch 1298/1500
Epoch 01297: val_loss did not improve
3s - loss: 5.8696e-04 - val_loss: 5.7987e-04
Epoch 1299/1500
Epoch 01298: val_loss did not improve
3s - loss: 5.9388e-04 - val_loss: 5.8684e-04
Epoch 1300/1500
Epoch 01299: val_loss did not improve
3s - loss: 5.7244e-04 - val_loss: 5.6938e-04
Epoch 1301/1500
Epoch 01300: val_loss did not improve
3s - loss: 5.8899e-04 - val_loss: 5.5616e-04
Epoch 1302/1500
Epoch 01301: val_loss did not improve
3s - loss: 5.9980e-04 - val_loss: 5.7570e-04
Epoch 1303/1500
Epoch 01302: val_loss did not improve
3s - loss: 5.7501e-04 - val_loss: 5.5536e-04
Epoch 1304/1500
Epoch 01303: val_loss did not improve
3s - loss: 5.8627e-04 - val_loss: 5.8837e-04
Epoch 1305/1500
Epoch 01304: val_loss did not improve
3s - loss: 5.8645e-04 - val_loss: 5.5998e-04
Epoch 1306/1500
Epoch 01305: val_loss did not improve
3s - loss: 5.7334e-04 - val_loss: 5.5321e-04
Epoch 1307/1500
Epoch 01306: val_loss did not improve
3s - loss: 5.8324e-04 - val_loss: 5.7805e-04
Epoch 1308/1500
Epoch 01307: val_loss did not improve
3s - loss: 5.6015e-04 - val_loss: 5.7228e-04
Epoch 1309/1500
Epoch 01308: val_loss did not improve
3s - loss: 5.6905e-04 - val_loss: 5.6183e-04
Epoch 1310/1500
Epoch 01309: val_loss did not improve
3s - loss: 5.9477e-04 - val_loss: 5.8364e-04
Epoch 1311/1500
Epoch 01310: val_loss did not improve
3s - loss: 5.8075e-04 - val_loss: 5.9057e-04
Epoch 1312/1500
Epoch 01311: val_loss did not improve
3s - loss: 5.8600e-04 - val_loss: 5.6894e-04
Epoch 1313/1500
Epoch 01312: val_loss did not improve
3s - loss: 5.8351e-04 - val_loss: 5.4472e-04
Epoch 1314/1500
Epoch 01313: val_loss did not improve
3s - loss: 5.7159e-04 - val_loss: 5.8957e-04
Epoch 1315/1500
Epoch 01314: val_loss did not improve
3s - loss: 5.7809e-04 - val_loss: 5.8107e-04
Epoch 1316/1500
Epoch 01315: val_loss did not improve
3s - loss: 5.7032e-04 - val_loss: 5.5177e-04
Epoch 1317/1500
Epoch 01316: val_loss did not improve
3s - loss: 5.8556e-04 - val_loss: 5.6444e-04
Epoch 1318/1500
Epoch 01317: val_loss did not improve
3s - loss: 5.7881e-04 - val_loss: 5.6006e-04
Epoch 1319/1500
Epoch 01318: val_loss did not improve
3s - loss: 5.6623e-04 - val_loss: 5.7524e-04
Epoch 1320/1500
Epoch 01319: val_loss did not improve
3s - loss: 5.8424e-04 - val_loss: 5.7888e-04
Epoch 1321/1500
Epoch 01320: val_loss did not improve
3s - loss: 5.8799e-04 - val_loss: 5.5969e-04
Epoch 1322/1500
Epoch 01321: val_loss did not improve
3s - loss: 5.9859e-04 - val_loss: 5.8266e-04
Epoch 1323/1500
Epoch 01322: val_loss did not improve
3s - loss: 5.6742e-04 - val_loss: 5.7141e-04
Epoch 1324/1500
Epoch 01323: val_loss did not improve
3s - loss: 5.6332e-04 - val_loss: 5.8142e-04
Epoch 1325/1500
Epoch 01324: val_loss did not improve
3s - loss: 5.7287e-04 - val_loss: 5.9239e-04
Epoch 1326/1500
Epoch 01325: val_loss did not improve
3s - loss: 5.8014e-04 - val_loss: 5.6553e-04
Epoch 1327/1500
Epoch 01326: val_loss did not improve
3s - loss: 5.7405e-04 - val_loss: 5.7644e-04
Epoch 1328/1500
Epoch 01327: val_loss did not improve
3s - loss: 5.5881e-04 - val_loss: 5.7680e-04
Epoch 1329/1500
Epoch 01328: val_loss did not improve
3s - loss: 5.9323e-04 - val_loss: 5.5729e-04
Epoch 1330/1500
Epoch 01329: val_loss did not improve
3s - loss: 5.8305e-04 - val_loss: 5.4775e-04
Epoch 1331/1500
Epoch 01330: val_loss did not improve
3s - loss: 5.7845e-04 - val_loss: 5.6833e-04
Epoch 1332/1500
Epoch 01331: val_loss did not improve
3s - loss: 5.6830e-04 - val_loss: 5.6672e-04
Epoch 1333/1500
Epoch 01332: val_loss did not improve
3s - loss: 5.6524e-04 - val_loss: 5.7576e-04
Epoch 1334/1500
Epoch 01333: val_loss did not improve
3s - loss: 5.9187e-04 - val_loss: 5.5824e-04
Epoch 1335/1500
Epoch 01334: val_loss did not improve
3s - loss: 6.0120e-04 - val_loss: 5.5649e-04
Epoch 1336/1500
Epoch 01335: val_loss did not improve
3s - loss: 6.0725e-04 - val_loss: 5.5744e-04
Epoch 1337/1500
Epoch 01336: val_loss did not improve
3s - loss: 5.4335e-04 - val_loss: 5.6424e-04
Epoch 1338/1500
Epoch 01337: val_loss did not improve
3s - loss: 5.8170e-04 - val_loss: 5.5202e-04
Epoch 1339/1500
Epoch 01338: val_loss did not improve
3s - loss: 5.9709e-04 - val_loss: 5.6135e-04
Epoch 1340/1500
Epoch 01339: val_loss did not improve
3s - loss: 5.9114e-04 - val_loss: 5.6201e-04
Epoch 1341/1500
Epoch 01340: val_loss did not improve
3s - loss: 5.7719e-04 - val_loss: 5.6174e-04
Epoch 1342/1500
Epoch 01341: val_loss did not improve
3s - loss: 5.9692e-04 - val_loss: 5.5956e-04
Epoch 1343/1500
Epoch 01342: val_loss did not improve
3s - loss: 5.9212e-04 - val_loss: 5.7346e-04
Epoch 1344/1500
Epoch 01343: val_loss did not improve
3s - loss: 5.6298e-04 - val_loss: 5.5563e-04
Epoch 1345/1500
Epoch 01344: val_loss did not improve
3s - loss: 5.6815e-04 - val_loss: 5.9190e-04
Epoch 1346/1500
Epoch 01345: val_loss did not improve
3s - loss: 5.8052e-04 - val_loss: 5.5520e-04
Epoch 1347/1500
Epoch 01346: val_loss did not improve
3s - loss: 5.7913e-04 - val_loss: 5.7967e-04
Epoch 1348/1500
Epoch 01347: val_loss did not improve
3s - loss: 5.7811e-04 - val_loss: 5.6994e-04
Epoch 1349/1500
Epoch 01348: val_loss did not improve
3s - loss: 5.6589e-04 - val_loss: 5.8458e-04
Epoch 1350/1500
Epoch 01349: val_loss did not improve
3s - loss: 5.8564e-04 - val_loss: 5.8026e-04
Epoch 1351/1500
Epoch 01350: val_loss did not improve
3s - loss: 5.7743e-04 - val_loss: 5.7949e-04
Epoch 1352/1500
Epoch 01351: val_loss did not improve
3s - loss: 5.8684e-04 - val_loss: 5.5810e-04
Epoch 1353/1500
Epoch 01352: val_loss did not improve
3s - loss: 5.8990e-04 - val_loss: 5.5616e-04
Epoch 1354/1500
Epoch 01353: val_loss did not improve
3s - loss: 5.8184e-04 - val_loss: 5.4870e-04
Epoch 1355/1500
Epoch 01354: val_loss did not improve
3s - loss: 6.0147e-04 - val_loss: 5.9273e-04
Epoch 1356/1500
Epoch 01355: val_loss did not improve
3s - loss: 5.8995e-04 - val_loss: 5.7469e-04
Epoch 1357/1500
Epoch 01356: val_loss did not improve
3s - loss: 5.7564e-04 - val_loss: 5.6515e-04
Epoch 1358/1500
Epoch 01357: val_loss did not improve
3s - loss: 5.7398e-04 - val_loss: 5.5817e-04
Epoch 1359/1500
Epoch 01358: val_loss did not improve
3s - loss: 5.5975e-04 - val_loss: 5.6929e-04
Epoch 1360/1500
Epoch 01359: val_loss did not improve
3s - loss: 6.0077e-04 - val_loss: 5.7267e-04
Epoch 1361/1500
Epoch 01360: val_loss did not improve
3s - loss: 5.7885e-04 - val_loss: 5.6858e-04
Epoch 1362/1500
Epoch 01361: val_loss did not improve
3s - loss: 5.7586e-04 - val_loss: 5.7693e-04
Epoch 1363/1500
Epoch 01362: val_loss did not improve
3s - loss: 5.7396e-04 - val_loss: 5.7753e-04
Epoch 1364/1500
Epoch 01363: val_loss did not improve
3s - loss: 5.8818e-04 - val_loss: 5.8329e-04
Epoch 1365/1500
Epoch 01364: val_loss did not improve
3s - loss: 5.9464e-04 - val_loss: 5.6351e-04
Epoch 1366/1500
Epoch 01365: val_loss did not improve
3s - loss: 5.9954e-04 - val_loss: 5.8165e-04
Epoch 1367/1500
Epoch 01366: val_loss did not improve
3s - loss: 5.9001e-04 - val_loss: 5.7208e-04
Epoch 1368/1500
Epoch 01367: val_loss did not improve
3s - loss: 5.8112e-04 - val_loss: 5.5063e-04
Epoch 1369/1500
Epoch 01368: val_loss did not improve
3s - loss: 5.7184e-04 - val_loss: 5.7256e-04
Epoch 1370/1500
Epoch 01369: val_loss did not improve
3s - loss: 5.7729e-04 - val_loss: 5.6955e-04
Epoch 1371/1500
Epoch 01370: val_loss did not improve
3s - loss: 5.8576e-04 - val_loss: 5.6883e-04
Epoch 1372/1500
Epoch 01371: val_loss did not improve
3s - loss: 5.6028e-04 - val_loss: 6.0565e-04
Epoch 1373/1500
Epoch 01372: val_loss did not improve
3s - loss: 5.7069e-04 - val_loss: 5.6981e-04
Epoch 1374/1500
Epoch 01373: val_loss did not improve
3s - loss: 5.7982e-04 - val_loss: 5.7101e-04
Epoch 1375/1500
Epoch 01374: val_loss did not improve
3s - loss: 5.7261e-04 - val_loss: 5.6623e-04
Epoch 1376/1500
Epoch 01375: val_loss did not improve
3s - loss: 5.8806e-04 - val_loss: 5.6373e-04
Epoch 1377/1500
Epoch 01376: val_loss did not improve
3s - loss: 5.8084e-04 - val_loss: 5.6116e-04
Epoch 1378/1500
Epoch 01377: val_loss did not improve
3s - loss: 5.7592e-04 - val_loss: 5.6536e-04
Epoch 1379/1500
Epoch 01378: val_loss did not improve
3s - loss: 5.7633e-04 - val_loss: 5.7334e-04
Epoch 1380/1500
Epoch 01379: val_loss did not improve
3s - loss: 5.8717e-04 - val_loss: 5.7771e-04
Epoch 1381/1500
Epoch 01380: val_loss did not improve
3s - loss: 5.7900e-04 - val_loss: 5.6786e-04
Epoch 1382/1500
Epoch 01381: val_loss did not improve
3s - loss: 5.8021e-04 - val_loss: 5.6986e-04
Epoch 1383/1500
Epoch 01382: val_loss did not improve
3s - loss: 5.6612e-04 - val_loss: 5.6540e-04
Epoch 1384/1500
Epoch 01383: val_loss did not improve
3s - loss: 5.9252e-04 - val_loss: 5.9234e-04
Epoch 1385/1500
Epoch 01384: val_loss did not improve
3s - loss: 6.0091e-04 - val_loss: 5.5682e-04
Epoch 1386/1500
Epoch 01385: val_loss did not improve
3s - loss: 5.9124e-04 - val_loss: 5.7895e-04
Epoch 1387/1500
Epoch 01386: val_loss did not improve
3s - loss: 5.6347e-04 - val_loss: 5.7157e-04
Epoch 1388/1500
Epoch 01387: val_loss did not improve
3s - loss: 5.6678e-04 - val_loss: 5.6958e-04
Epoch 1389/1500
Epoch 01388: val_loss did not improve
3s - loss: 5.7837e-04 - val_loss: 5.6180e-04
Epoch 1390/1500
Epoch 01389: val_loss did not improve
3s - loss: 5.7711e-04 - val_loss: 5.8515e-04
Epoch 1391/1500
Epoch 01390: val_loss did not improve
3s - loss: 5.6707e-04 - val_loss: 5.8416e-04
Epoch 1392/1500
Epoch 01391: val_loss did not improve
3s - loss: 5.8210e-04 - val_loss: 5.5984e-04
Epoch 1393/1500
Epoch 01392: val_loss did not improve
3s - loss: 5.8546e-04 - val_loss: 5.6249e-04
Epoch 1394/1500
Epoch 01393: val_loss did not improve
3s - loss: 5.8558e-04 - val_loss: 5.6953e-04
Epoch 1395/1500
Epoch 01394: val_loss did not improve
3s - loss: 5.8338e-04 - val_loss: 5.8555e-04
Epoch 1396/1500
Epoch 01395: val_loss did not improve
3s - loss: 6.0331e-04 - val_loss: 5.7067e-04
Epoch 1397/1500
Epoch 01396: val_loss did not improve
3s - loss: 5.6253e-04 - val_loss: 5.7318e-04
Epoch 1398/1500
Epoch 01397: val_loss did not improve
3s - loss: 5.9023e-04 - val_loss: 5.9451e-04
Epoch 1399/1500
Epoch 01398: val_loss did not improve
3s - loss: 5.7847e-04 - val_loss: 6.2787e-04
Epoch 1400/1500
Epoch 01399: val_loss did not improve
3s - loss: 5.9174e-04 - val_loss: 5.8288e-04
Epoch 1401/1500
Epoch 01400: val_loss did not improve
3s - loss: 5.7727e-04 - val_loss: 5.4430e-04
Epoch 1402/1500
Epoch 01401: val_loss did not improve
3s - loss: 5.7913e-04 - val_loss: 5.5920e-04
Epoch 1403/1500
Epoch 01402: val_loss did not improve
3s - loss: 5.8426e-04 - val_loss: 5.8928e-04
Epoch 1404/1500
Epoch 01403: val_loss did not improve
3s - loss: 5.7542e-04 - val_loss: 5.8464e-04
Epoch 1405/1500
Epoch 01404: val_loss did not improve
3s - loss: 5.7842e-04 - val_loss: 5.7373e-04
Epoch 1406/1500
Epoch 01405: val_loss did not improve
3s - loss: 5.7778e-04 - val_loss: 5.5979e-04
Epoch 1407/1500
Epoch 01406: val_loss did not improve
3s - loss: 5.8918e-04 - val_loss: 5.8948e-04
Epoch 1408/1500
Epoch 01407: val_loss did not improve
3s - loss: 5.7079e-04 - val_loss: 5.6383e-04
Epoch 1409/1500
Epoch 01408: val_loss did not improve
3s - loss: 5.9608e-04 - val_loss: 5.7254e-04
Epoch 1410/1500
Epoch 01409: val_loss did not improve
3s - loss: 5.8771e-04 - val_loss: 5.8331e-04
Epoch 1411/1500
Epoch 01410: val_loss did not improve
3s - loss: 5.9467e-04 - val_loss: 5.9292e-04
Epoch 1412/1500
Epoch 01411: val_loss did not improve
3s - loss: 5.8670e-04 - val_loss: 5.8009e-04
Epoch 1413/1500
Epoch 01412: val_loss did not improve
3s - loss: 5.5094e-04 - val_loss: 5.8433e-04
Epoch 1414/1500
Epoch 01413: val_loss did not improve
3s - loss: 5.7772e-04 - val_loss: 5.6181e-04
Epoch 1415/1500
Epoch 01414: val_loss did not improve
3s - loss: 5.9241e-04 - val_loss: 5.9098e-04
Epoch 1416/1500
Epoch 01415: val_loss did not improve
3s - loss: 5.7808e-04 - val_loss: 6.0161e-04
Epoch 1417/1500
Epoch 01416: val_loss did not improve
3s - loss: 5.8068e-04 - val_loss: 5.7369e-04
Epoch 1418/1500
Epoch 01417: val_loss did not improve
3s - loss: 5.6597e-04 - val_loss: 5.5434e-04
Epoch 1419/1500
Epoch 01418: val_loss did not improve
3s - loss: 5.7081e-04 - val_loss: 6.1091e-04
Epoch 1420/1500
Epoch 01419: val_loss did not improve
3s - loss: 5.8332e-04 - val_loss: 5.7448e-04
Epoch 1421/1500
Epoch 01420: val_loss did not improve
3s - loss: 5.8273e-04 - val_loss: 5.7652e-04
Epoch 1422/1500
Epoch 01421: val_loss did not improve
3s - loss: 5.7225e-04 - val_loss: 5.7702e-04
Epoch 1423/1500
Epoch 01422: val_loss did not improve
3s - loss: 6.2281e-04 - val_loss: 5.7603e-04
Epoch 1424/1500
Epoch 01423: val_loss did not improve
3s - loss: 5.8311e-04 - val_loss: 5.5696e-04
Epoch 1425/1500
Epoch 01424: val_loss did not improve
3s - loss: 5.8155e-04 - val_loss: 5.7942e-04
Epoch 1426/1500
Epoch 01425: val_loss did not improve
3s - loss: 5.9365e-04 - val_loss: 5.8684e-04
Epoch 1427/1500
Epoch 01426: val_loss did not improve
3s - loss: 5.6672e-04 - val_loss: 5.5643e-04
Epoch 1428/1500
Epoch 01427: val_loss did not improve
3s - loss: 5.6171e-04 - val_loss: 5.7793e-04
Epoch 1429/1500
Epoch 01428: val_loss did not improve
3s - loss: 5.6314e-04 - val_loss: 5.6839e-04
Epoch 1430/1500
Epoch 01429: val_loss did not improve
3s - loss: 5.8173e-04 - val_loss: 6.0898e-04
Epoch 1431/1500
Epoch 01430: val_loss did not improve
3s - loss: 5.8959e-04 - val_loss: 5.7279e-04
Epoch 1432/1500
Epoch 01431: val_loss did not improve
3s - loss: 5.6723e-04 - val_loss: 5.7019e-04
Epoch 1433/1500
Epoch 01432: val_loss did not improve
3s - loss: 5.6235e-04 - val_loss: 5.7655e-04
Epoch 1434/1500
Epoch 01433: val_loss did not improve
3s - loss: 5.6922e-04 - val_loss: 5.8674e-04
Epoch 1435/1500
Epoch 01434: val_loss did not improve
3s - loss: 5.9083e-04 - val_loss: 6.1962e-04
Epoch 1436/1500
Epoch 01435: val_loss did not improve
3s - loss: 5.8518e-04 - val_loss: 5.6167e-04
Epoch 1437/1500
Epoch 01436: val_loss did not improve
3s - loss: 5.6528e-04 - val_loss: 5.7917e-04
Epoch 1438/1500
Epoch 01437: val_loss did not improve
3s - loss: 5.8033e-04 - val_loss: 5.7800e-04
Epoch 1439/1500
Epoch 01438: val_loss did not improve
3s - loss: 5.6909e-04 - val_loss: 5.7167e-04
Epoch 1440/1500
Epoch 01439: val_loss did not improve
3s - loss: 5.8341e-04 - val_loss: 5.6549e-04
Epoch 1441/1500
Epoch 01440: val_loss did not improve
3s - loss: 5.7681e-04 - val_loss: 5.6788e-04
Epoch 1442/1500
Epoch 01441: val_loss did not improve
3s - loss: 5.8338e-04 - val_loss: 5.6272e-04
Epoch 1443/1500
Epoch 01442: val_loss did not improve
3s - loss: 5.6698e-04 - val_loss: 5.6229e-04
Epoch 1444/1500
Epoch 01443: val_loss did not improve
3s - loss: 5.7091e-04 - val_loss: 5.9311e-04
Epoch 1445/1500
Epoch 01444: val_loss did not improve
3s - loss: 5.5675e-04 - val_loss: 5.6062e-04
Epoch 1446/1500
Epoch 01445: val_loss did not improve
3s - loss: 5.7962e-04 - val_loss: 5.4686e-04
Epoch 1447/1500
Epoch 01446: val_loss did not improve
3s - loss: 5.6360e-04 - val_loss: 5.7272e-04
Epoch 1448/1500
Epoch 01447: val_loss did not improve
3s - loss: 5.6292e-04 - val_loss: 5.5466e-04
Epoch 1449/1500
Epoch 01448: val_loss did not improve
3s - loss: 5.6813e-04 - val_loss: 5.4832e-04
Epoch 1450/1500
Epoch 01449: val_loss did not improve
3s - loss: 5.8560e-04 - val_loss: 5.7321e-04
Epoch 1451/1500
Epoch 01450: val_loss did not improve
3s - loss: 5.6392e-04 - val_loss: 5.6764e-04
Epoch 1452/1500
Epoch 01451: val_loss did not improve
3s - loss: 5.6082e-04 - val_loss: 5.7044e-04
Epoch 1453/1500
Epoch 01452: val_loss did not improve
3s - loss: 5.7483e-04 - val_loss: 5.3651e-04
Epoch 1454/1500
Epoch 01453: val_loss did not improve
3s - loss: 5.6946e-04 - val_loss: 5.6176e-04
Epoch 1455/1500
Epoch 01454: val_loss did not improve
3s - loss: 5.6990e-04 - val_loss: 5.7772e-04
Epoch 1456/1500
Epoch 01455: val_loss did not improve
3s - loss: 5.4525e-04 - val_loss: 5.7212e-04
Epoch 1457/1500
Epoch 01456: val_loss did not improve
3s - loss: 5.6248e-04 - val_loss: 5.7761e-04
Epoch 1458/1500
Epoch 01457: val_loss did not improve
3s - loss: 5.8838e-04 - val_loss: 5.6519e-04
Epoch 1459/1500
Epoch 01458: val_loss did not improve
3s - loss: 5.6178e-04 - val_loss: 5.8135e-04
Epoch 1460/1500
Epoch 01459: val_loss did not improve
3s - loss: 5.8506e-04 - val_loss: 5.5561e-04
Epoch 1461/1500
Epoch 01460: val_loss did not improve
3s - loss: 5.7569e-04 - val_loss: 5.6884e-04
Epoch 1462/1500
Epoch 01461: val_loss did not improve
3s - loss: 5.6962e-04 - val_loss: 5.6954e-04
Epoch 1463/1500
Epoch 01462: val_loss did not improve
3s - loss: 5.6743e-04 - val_loss: 5.7246e-04
Epoch 1464/1500
Epoch 01463: val_loss did not improve
3s - loss: 5.6810e-04 - val_loss: 5.7092e-04
Epoch 1465/1500
Epoch 01464: val_loss did not improve
3s - loss: 5.7425e-04 - val_loss: 5.6789e-04
Epoch 1466/1500
Epoch 01465: val_loss did not improve
3s - loss: 5.8324e-04 - val_loss: 5.5548e-04
Epoch 1467/1500
Epoch 01466: val_loss did not improve
3s - loss: 5.8901e-04 - val_loss: 5.6671e-04
Epoch 1468/1500
Epoch 01467: val_loss did not improve
3s - loss: 5.6283e-04 - val_loss: 5.9557e-04
Epoch 1469/1500
Epoch 01468: val_loss did not improve
3s - loss: 5.6796e-04 - val_loss: 5.8370e-04
Epoch 1470/1500
Epoch 01469: val_loss did not improve
3s - loss: 5.6393e-04 - val_loss: 5.6933e-04
Epoch 1471/1500
Epoch 01470: val_loss did not improve
3s - loss: 5.7109e-04 - val_loss: 5.7584e-04
Epoch 1472/1500
Epoch 01471: val_loss did not improve
3s - loss: 5.4949e-04 - val_loss: 5.7588e-04
Epoch 1473/1500
Epoch 01472: val_loss did not improve
3s - loss: 5.9253e-04 - val_loss: 5.6241e-04
Epoch 1474/1500
Epoch 01473: val_loss did not improve
3s - loss: 5.6513e-04 - val_loss: 5.9284e-04
Epoch 1475/1500
Epoch 01474: val_loss did not improve
3s - loss: 5.7764e-04 - val_loss: 6.3485e-04
Epoch 1476/1500
Epoch 01475: val_loss did not improve
3s - loss: 5.8154e-04 - val_loss: 5.6494e-04
Epoch 1477/1500
Epoch 01476: val_loss did not improve
3s - loss: 5.7851e-04 - val_loss: 5.9405e-04
Epoch 1478/1500
Epoch 01477: val_loss did not improve
3s - loss: 5.6407e-04 - val_loss: 6.0223e-04
Epoch 1479/1500
Epoch 01478: val_loss did not improve
3s - loss: 5.8636e-04 - val_loss: 5.9635e-04
Epoch 1480/1500
Epoch 01479: val_loss did not improve
3s - loss: 5.6187e-04 - val_loss: 5.6960e-04
Epoch 1481/1500
Epoch 01480: val_loss did not improve
3s - loss: 5.7533e-04 - val_loss: 5.5613e-04
Epoch 1482/1500
Epoch 01481: val_loss did not improve
3s - loss: 5.7787e-04 - val_loss: 5.9883e-04
Epoch 1483/1500
Epoch 01482: val_loss did not improve
3s - loss: 5.2598e-04 - val_loss: 5.8297e-04
Epoch 1484/1500
Epoch 01483: val_loss did not improve
3s - loss: 5.6093e-04 - val_loss: 5.7229e-04
Epoch 1485/1500
Epoch 01484: val_loss did not improve
3s - loss: 5.7281e-04 - val_loss: 5.9565e-04
Epoch 1486/1500
Epoch 01485: val_loss did not improve
3s - loss: 5.5933e-04 - val_loss: 5.8750e-04
Epoch 1487/1500
Epoch 01486: val_loss did not improve
3s - loss: 5.4925e-04 - val_loss: 5.6611e-04
Epoch 1488/1500
Epoch 01487: val_loss did not improve
3s - loss: 5.8204e-04 - val_loss: 5.7734e-04
Epoch 1489/1500
Epoch 01488: val_loss did not improve
3s - loss: 5.5913e-04 - val_loss: 5.7450e-04
Epoch 1490/1500
Epoch 01489: val_loss did not improve
3s - loss: 5.6948e-04 - val_loss: 6.0066e-04
Epoch 1491/1500
Epoch 01490: val_loss did not improve
3s - loss: 5.6808e-04 - val_loss: 5.9377e-04
Epoch 1492/1500
Epoch 01491: val_loss did not improve
3s - loss: 5.7451e-04 - val_loss: 5.9423e-04
Epoch 1493/1500
Epoch 01492: val_loss did not improve
3s - loss: 5.7978e-04 - val_loss: 5.8075e-04
Epoch 1494/1500
Epoch 01493: val_loss did not improve
3s - loss: 5.6753e-04 - val_loss: 5.6953e-04
Epoch 1495/1500
Epoch 01494: val_loss did not improve
3s - loss: 5.7053e-04 - val_loss: 5.7777e-04
Epoch 1496/1500
Epoch 01495: val_loss did not improve
3s - loss: 5.7423e-04 - val_loss: 5.7161e-04
Epoch 1497/1500
Epoch 01496: val_loss did not improve
3s - loss: 5.7490e-04 - val_loss: 5.7208e-04
Epoch 1498/1500
Epoch 01497: val_loss did not improve
3s - loss: 5.7103e-04 - val_loss: 5.6293e-04
Epoch 1499/1500
Epoch 01498: val_loss did not improve
3s - loss: 5.7123e-04 - val_loss: 5.6973e-04
Epoch 1500/1500
Epoch 01499: val_loss did not improve
3s - loss: 5.7753e-04 - val_loss: 5.8073e-04

---

Step 7: Visualize the Loss and Test Predictions

(IMPLEMENTATION) Answer a few questions and visualize the loss

Question 1: Outline the steps you took to get to your final neural network architecture and your reasoning at each step.

Answer: I started with a simple CNN architecture: 2 convolutional layers with 32 and 64 filters interleaved with pooling layers, one fully-connected layer and one output layer with the size equal to the number of X and Y keypoint coordinates (2 * 15 = 30). This is a regression problem so, obviously, the loss function must be 'mean_squared_error'. To combat overfitting, I used dropout layers. With the initial architecture, I got the validation accuracy around 0.0015. Then I was experimenting with the number of convolutional and fully-connected layers, dropout rate, mini batch size and the number of epochs. My final no-augmentation CNN architecture is shown above in cell 25: modelNoAugm. With that model, I got the validation accuracy equal to 0.00064. In order to improve it even more, I augmented the training data by applying horizontal flip transformation to the original training images, as it was suggested in the blog. Flipped images are obtained on the fly by transforming 50% of the batch images in the function generator_train. To avoid overfitting on the increased number of training images I slightly increased dropout rates of some CNN layers. With data augmentation, the validation loss improved to 0.000534 and this validation accuracy seems to be very good compared to the accuracy values obtained by the author of the blog.

Question 2: Defend your choice of optimizer. Which optimizers did you test, and how did you determine which worked best?

Answer: I tested adam, adadelta and rmsprop. I got the best results using 'adam' with the default parameters. With the RMSprop optimizer the CNN was not converging well and after 1000 epoch the loss and val_loss values were still around 0.0044. Maybe this particular CNN architecture is not good for the RMSProp optmizer or I should have used different parameters. I did not investigate it much because I got excellent results with 'adam'. The 'adadelta' was my number one choice after it proved to be very effective and worked very well for my dog project's CNN. Unfortunately, for this regression problem it did not work well: CNN models converged very slowly, the best val. loss value I got was something around 0.001 after 1000 epochs. I spent a lot of time tweaking the CNN architecture but finally realized that the problem could be in the optimizer.

Use the code cell below to plot the training and validation loss of your neural network. You may find this resource useful.

## TODO: Visualize the training and validation loss of your neural network
import matplotlib.pyplot as plt
%matplotlib inline

# list all data in history
print(hist.history.keys())

# summarize history for loss
plt.plot(hist.history['loss'])
plt.plot(hist.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.ylim(1e-4, 1e-2)
plt.legend(['loss', 'val_loss'], loc='upper left')
plt.show()

dict_keys(['val_loss', 'loss'])

Question 3: Do you notice any evidence of overfitting or underfitting in the above plot? If so, what steps have you taken to improve your model? Note that slight overfitting or underfitting will not hurt your chances of a successful submission, as long as you have attempted some solutions towards improving your model (such as regularization, dropout, increased/decreased number of layers, etc).

Answer: In my model overfitting happens only at the very end of the training process. I used dropout layers to combat overfitting in my CNN.

Visualize a Subset of the Test Predictions

Execute the code cell below to visualize your model's predicted keypoints on a subset of the testing images.

y_test = model.predict(X_test)
fig = plt.figure(figsize=(20,20))
fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)
for i in range(9):
    ax = fig.add_subplot(3, 3, i + 1, xticks=[], yticks=[])
    plot_data(X_test[i], y_test[i], ax)

---

Step 8: Complete the pipeline

With the work you did in Sections 1 and 2 of this notebook, along with your freshly trained facial keypoint detector, you can now complete the full pipeline. That is given a color image containing a person or persons you can now

• Detect the faces in this image automatically using OpenCV
• Predict the facial keypoints in each face detected in the image
• Paint predicted keypoints on each face detected

In this Subsection you will do just this!

(IMPLEMENTATION) Facial Keypoints Detector

Use the OpenCV face detection functionality you built in previous Sections to expand the functionality of your keypoints detector to color images with arbitrary size. Your function should perform the following steps

1. Accept a color image.
2. Convert the image to grayscale.
3. Detect and crop the face contained in the image.
4. Locate the facial keypoints in the cropped image.
5. Overlay the facial keypoints in the original (color, uncropped) image.

Note: step 4 can be the trickiest because remember your convolutional network is only trained to detect facial keypoints in $96 \times 96$ grayscale images where each pixel was normalized to lie in the interval $[0,1]$, and remember that each facial keypoint was normalized during training to the interval $[-1,1]$. This means - practically speaking - to paint detected keypoints onto a test face you need to perform this same pre-processing to your candidate face - that is after detecting it you should resize it to $96 \times 96$ and normalize its values before feeding it into your facial keypoint detector. To be shown correctly on the original image the output keypoints from your detector then need to be shifted and re-normalized from the interval $[-1,1]$ to the width and height of your detected face.

When complete you should be able to produce example images like the one below

# Load in color image for face detection
image = cv2.imread('images/obamas4.jpg')


# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)


# plot our image
fig = plt.figure(figsize = (9,9))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('image')
ax1.imshow(image)

<matplotlib.image.AxesImage at 0x1f18bb1e438>

### TODO: Use the face detection code we saw in Section 1 with your trained conv-net 

# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 1.25, 6)

image_copy = np.copy(image)

coord_X = list()
coord_Y = list()
# Get the bounding box for each detected face
for (x,y,w,h) in faces:
 
    # Crop the face image
    face_image = gray[y:y+h, x:x+w]
    face_image = cv2.resize(face_image, (96, 96), interpolation=cv2.INTER_CUBIC)

    # Scale pixels and reshape for prediction
    X = face_image / 255.
    X = X.astype(np.float32)
    X = X.reshape(-1, 96, 96, 1)
    
    # Predict keypoints
    res = model.predict(X)
    
    # Get keypoint coordinates on the original image
    res_X = np.squeeze(res)[0::2] * w / 2 + w / 2 + x
    res_Y = np.squeeze(res)[1::2] * h / 2 + h / 2 + y  
    
    # Save the coordinates 
    coord_X.append(np.array(res_X))
    coord_Y.append(np.array(res_Y))
    
    # Add a red bounding box to the detections image
    cv2.rectangle(image_copy, (x,y), (x+w,y+h), (255,0,0), 3)
    

    
# Plot our image
fig = plt.figure(figsize = (9,9))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('image copy')
ax1.imshow(image_copy)

## TODO : Paint the predicted keypoints on the test image
keypoints_X = np.array(coord_X)
keypoints_Y = np.array(coord_Y)
ax1.scatter(keypoints_X, keypoints_Y, marker='o', c='c', s=5)

<matplotlib.collections.PathCollection at 0x1f18218d160>

(Optional) Further Directions - add a filter using facial keypoints to your laptop camera

Now you can add facial keypoint detection to your laptop camera - as illustrated in the gif below.

The next Python cell contains the basic laptop video camera function used in the previous optional video exercises. Combine it with the functionality you developed for keypoint detection and marking in the previous exercise and you should be good to go!

import cv2
import time 
from keras.models import load_model
def laptop_camera_go():
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Extract the pre-trained face detector from an xml file
    face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

    # Define the codec and create VideoWriter object
    fourcc = cv2.VideoWriter_fourcc(*'XVID')
    out = cv2.VideoWriter('output.avi',fourcc, 10.0, (640,480))
    
    # keep video stream open
    while rval:
        # Convert the RGB  image to grayscale
        gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)

        # Detect the faces in image
        faces = face_cascade.detectMultiScale(gray, 1.1, 4)

        image_copy = np.copy(frame)

        # Get the bounding box for each detected face
        for (x,y,w,h) in faces:
 
            # Crop the face image
            face_image = gray[y:y+h, x:x+w]
            face_image = cv2.resize(face_image, (96, 96), interpolation=cv2.INTER_CUBIC)

            # Scale pixels and reshape for prediction
            X = face_image / 255.
            X = X.astype(np.float32)
            X = X.reshape(-1, 96, 96, 1)
    
            # Predict keypoints
            res = model.predict(X)
    
            # Get keypoint coordinates on the original image
            res_X = np.squeeze(res)[0::2] * w / 2 + w / 2 + x
            res_Y = np.squeeze(res)[1::2] * h / 2 + h / 2 + y  
    
            for i, c_x in enumerate(res_X):
                cv2.circle(image_copy, (int(c_x), int(res_Y[i])), 2, (0,255,0), -1)
        
            # Add a red bounding box to the detections image
            cv2.rectangle(image_copy, (x,y), (x+w,y+h), (255,0,0), 3)
            
        # write the frame
        out.write(image_copy)
            
        # plot image from camera with detections marked
        cv2.imshow("face detection activated", image_copy)
        
        # exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # exit by pressing any key
            vc.release()
            out.release()

            # destroy windows
            cv2.destroyAllWindows()
            
            # hack from stack overflow for making sure window closes on osx --> https://stackoverflow.com/questions/6116564/destroywindow-does-not-close-window-on-mac-using-python-and-opencv
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()  

# Run your keypoint face painter
laptop_camera_go()

(Optional) Further Directions - add a filter using facial keypoints

Using your freshly minted facial keypoint detector pipeline you can now do things like add fun filters to a person's face automatically. In this optional exercise you can play around with adding sunglasses automatically to each individual's face in an image as shown in a demonstration image below.

To produce this effect an image of a pair of sunglasses shown in the Python cell below.

# Load in sunglasses image - note the usage of the special option
# cv2.IMREAD_UNCHANGED, this option is used because the sunglasses 
# image has a 4th channel that allows us to control how transparent each pixel in the image is
sunglasses = cv2.imread("images/sunglasses_4.png", cv2.IMREAD_UNCHANGED)

# Plot the image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.imshow(sunglasses)
ax1.axis('off');

This image is placed over each individual's face using the detected eye points to determine the location of the sunglasses, and eyebrow points to determine the size that the sunglasses should be for each person (one could also use the nose point to determine this).

Notice that this image actually has 4 channels, not just 3.

# Print out the shape of the sunglasses image
print ('The sunglasses image has shape: ' + str(np.shape(sunglasses)))

The sunglasses image has shape: (1123, 3064, 4)

It has the usual red, blue, and green channels any color image has, with the 4th channel representing the transparency level of each pixel in the image. Here's how the transparency channel works: the lower the value, the more transparent the pixel will become. The lower bound (completely transparent) is zero here, so any pixels set to 0 will not be seen.

This is how we can place this image of sunglasses on someone's face and still see the area around of their face where the sunglasses lie - because these pixels in the sunglasses image have been made completely transparent.

Lets check out the alpha channel of our sunglasses image in the next Python cell. Note because many of the pixels near the boundary are transparent we'll need to explicitly print out non-zero values if we want to see them.

# Print out the sunglasses transparency (alpha) channel
alpha_channel = sunglasses[:,:,3]
print ('the alpha channel here looks like')
print (alpha_channel)

# Just to double check that there are indeed non-zero values
# Let's find and print out every value greater than zero
values = np.where(alpha_channel != 0)
print ('\n the non-zero values of the alpha channel look like')
print (values)

the alpha channel here looks like
[[0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 ..., 
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]]

 the non-zero values of the alpha channel look like
(array([  17,   17,   17, ..., 1109, 1109, 1109], dtype=int64), array([ 687,  688,  689, ..., 2376, 2377, 2378], dtype=int64))

This means that when we place this sunglasses image on top of another image, we can use the transparency channel as a filter to tell us which pixels to overlay on a new image (only the non-transparent ones with values greater than zero).

One last thing: it's helpful to understand which keypoint belongs to the eyes, mouth, etc. So, in the image below, we also display the index of each facial keypoint directly on the image so that you can tell which keypoints are for the eyes, eyebrows, etc.

With this information, you're well on your way to completing this filtering task! See if you can place the sunglasses automatically on the individuals in the image loaded in / shown in the next Python cell.

# Load in color image for face detection
image = cv2.imread('images/obamas4.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)


# Plot the image
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('Original Image')
ax1.imshow(image)

<matplotlib.image.AxesImage at 0x1f18224e748>

## (Optional) TODO: Use the face detection code we saw in Section 1 with your trained conv-net to put
## sunglasses on the individuals in our test image

# Load in color image for face detection
image = cv2.imread('images/obamas4.jpg')

sunglasses = cv2.imread("images/sunglasses_4.png", cv2.IMREAD_UNCHANGED)

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 1.25, 6)

image_copy = np.copy(image)

# Glasses scale value
SCALE_COEFF = 1.15

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
 
    # Crop the face image
    face_image = gray[y:y+h, x:x+w]
    face_image = cv2.resize(face_image, (96, 96), interpolation=cv2.INTER_CUBIC)

    # Scale pixels and reshape for prediction
    X = face_image / 255.
    X = X.astype(np.float32)
    X = X.reshape(-1, 96, 96, 1)
    
    # Predict keypoints
    res = model.predict(X)
    
    # Get keypoint coordinates on the original image
    res_X = np.squeeze(res)[0::2] * w / 2 + w / 2 + x
    res_Y = np.squeeze(res)[1::2] * h / 2 + h / 2 + y  
    
    # Calculate the size of the glasses using the facial keypoints that belong to eyes
    glasses_width = int((res_X[7] - res_X[9]) * SCALE_COEFF)
    glasses_height = 2 * int(SCALE_COEFF * (max(res_Y[5], res_Y[1], res_Y[4]) - min(res_Y[9], res_Y[8])))
    
    scaled_glasses = cv2.resize(sunglasses, (glasses_width, glasses_height), interpolation=cv2.INTER_CUBIC)

    # overlay the glasses image on the original image
    y_t = int(min(res_Y[9], res_Y[8]))
    y_b = y_t + glasses_height
    x_l = int((res_X[7] + res_X[9]) / 2 - glasses_width / 2)
    x_r = x_l + glasses_width
            
    alpha_glasses = scaled_glasses[:, :, 3] / 255.0
    alpha_image = 1.0 - alpha_glasses

    for c in range(0, 3):
        image_copy[y_t:y_b, x_l:x_r, c] = (alpha_glasses * scaled_glasses[:, :, c] +
                              alpha_image * image_copy[y_t:y_b, x_l:x_r, c])
    
# Plot our image
fig = plt.figure(figsize = (9,9))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('image copy')
ax1.imshow(image_copy)

<matplotlib.image.AxesImage at 0x1f182030e48>

(Optional) Further Directions - add a filter using facial keypoints to your laptop camera

Now you can add the sunglasses filter to your laptop camera - as illustrated in the gif below.

The next Python cell contains the basic laptop video camera function used in the previous optional video exercises. Combine it with the functionality you developed for adding sunglasses to someone's face in the previous optional exercise and you should be good to go!

import cv2
import time 
from keras.models import load_model
import numpy as np

# Glasses scale value
SCALE_COEFF = 1.15

def laptop_camera_go():
    sunglasses = cv2.imread("images/sunglasses_4.png", cv2.IMREAD_UNCHANGED)

    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Define the codec and create VideoWriter object
    fourcc = cv2.VideoWriter_fourcc(*'XVID')
    out = cv2.VideoWriter('output.avi',fourcc, 10.0, (640,480))
    
    # Keep video stream open
    while rval:
        # Convert the RGB  image to grayscale
        gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)

        # Detect the faces in image
        faces = face_cascade.detectMultiScale(gray, 1.1, 4)

        image_copy = np.copy(frame)

        # Get the bounding box for each detected face
        for (x,y,w,h) in faces:
 
            # Crop the face image
            face_image = gray[y:y+h, x:x+w]
            face_image = cv2.resize(face_image, (96, 96), interpolation=cv2.INTER_CUBIC)

            # Scale pixels and reshape for prediction
            X = face_image / 255.
            X = X.astype(np.float32)
            X = X.reshape(-1, 96, 96, 1)
    
            # Predict keypoints
            res = model.predict(X)
    
            # Get keypoint coordinates on the original image
            res_X = np.squeeze(res)[0::2] * w / 2 + w / 2 + x
            res_Y = np.squeeze(res)[1::2] * h / 2 + h / 2 + y  
    
             # Calculate the size of the glasses using points 7 and 9
            glasses_width = int((res_X[7] - res_X[9]) * SCALE_COEFF)
            glasses_height = 2 * int(SCALE_COEFF * (max(res_Y[5], res_Y[1], res_Y[4]) - min(res_Y[9], res_Y[8])))
    
            scaled_glasses = cv2.resize(sunglasses, (glasses_width, glasses_height), interpolation=cv2.INTER_CUBIC)

            # overlay the glasses image on the original image
            y_t = int(min(res_Y[9], res_Y[8]))
            y_b = y_t + glasses_height
            x_l = int((res_X[7] + res_X[9]) / 2 - glasses_width / 2)
            x_r = x_l + glasses_width

            alpha_glasses = scaled_glasses[:, :, 3] / 255.0
            alpha_image = 1.0 - alpha_glasses

            for c in range(0, 3):
                image_copy[y_t:y_b, x_l:x_r, c] = (alpha_glasses * scaled_glasses[:, :, c] +
                              alpha_image * image_copy[y_t:y_b, x_l:x_r, c])

        # write the frame
        out.write(image_copy)
            
        # plot image from camera with detections marked
        cv2.imshow("face detection activated", image_copy)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # exit by pressing any key
            vc.release()
            out.release()

            # Destroy windows 
            cv2.destroyAllWindows()
            
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    
        

# Load facial landmark detector model
model = load_model('my_model.h5')

# Run sunglasses painter
laptop_camera_go()